Introduction

Tobacco policy in the UK appears to have gone through a period of rapid change following decades of stability. In 2002 Westminster passed a bill banning tobacco advertising throughout the UK. In 2006 it passed a bill banning smoking comprehensively in enclosed public places in England. Similar decisions were made in Wales and Northern Ireland, while the Scottish Parliament was the first to legislate on the issue in 2005 (Cairney, forthcoming, 2007). This, combined with measures such as smoking cessation initiatives and health education, now makes the UK the most progressive member state in the EU on tobacco policy (see Joosens and Raw, 2006, http://tc.bmjjournals.com/cgi/content/full/15/3/24). Yet for most of the post-war period tobacco policy was marked by stability. Tobacco control measures were voluntary rather than legislative, while public health arguments often came second to those based on individual choice and the economic benefits of tobacco.

The aim of this article is to critically analyse these developments: to account for policy stability, to analyse the extent of change, and to explain these events using theoretical models of policy change. Of particular relevance are the advocacy coalition framework (ACF) and the punctuated equilibrium models, which seek to explain sudden change despite the presence of powerful policy networks. Other models — including policy transfer and multi-level governance — aid explanation by focusing on activity ‘above and below’ the UK, while multiple streams analysis points to idiosyncratic elements of change.

In this analysis, the article does not seek to confirm the value of one particular model. Rather, it constructs three arguments drawing on the idea of ‘narratives’ of policy explanation (see Bevir and Rhodes, 2003). First, apparent change does not mean actual change. The article challenges researchers to be careful about assuming the scope and nature of policy change before embarking on explanation. The discussion of measurement and the development of two alternative narratives highlight the difficulties involved in defining the scope and nature of policy change. Second, the selection of one narrative at the expense of another itself determines the value of these models. For example, models describing ‘external’ factors have more value if we demonstrate ‘internal’ inertia based on policy network dominance. If, instead, the network was open and conducive to long-term policy change, then external influences are less impressive. The third argument is that the ‘multiple lenses’ approach gives a more complete explanation of policy change. The article employs each model as a tool, to make them work for the case study rather than vice versa. As Allison (1969, 715–716) argues, the fact that different analysts relying on different models produce different explanations suggests the need for introspection. Policy analysts face the same problems as ‘rational’ decision-makers: an infinite wealth of potentially relevant information to choose from but finite resources with which to choose. As a result, both focus their attention on one particular aspect of explanation. Employing more than one model does not solve this problem, but it does highlight a series of perspectives through which to view the same phenomenon.

The article concludes with a discussion of generalizability. It suggests that while UK tobacco policy is broadly consistent with international policy change, the results may differ from other areas in British politics. The same can be said for most case studies since variation by policy area is a key tenet of public policy analysis (see John, 1998). Policy outcomes vary by the stage of the policy cycle, the level of government, and the policy sector. Policy analysis is therefore devoted to juggling parsimonious explanation with the acknowledgement of complexity. This article addresses both concerns by applying a multiple lenses approach to detailed case study analysis.

Policy Stability: A Smoking Policy Community?

Policy stability in post-war tobacco policy came from an insulated industry–government relationship underpinned by socio-economic conditions. However, the value of tobacco to government varies by source. For example, Action on Smoking and Health (ASH, 2005b) estimates that the tobacco industry provided 40,000 direct and indirect jobs in 1979, while the Tobacco Advisory Council (TAC) suggests a figure of 300,000 jobs, with many in economically depressed and/or marginal constituencies (Taylor, 1984, 69). In 1981 tobacco generated £4 billion in taxes (equivalent to the £9.3 billion raised in 2003), provided a successful industrial export, the prevalence of smoking was high (despite falling from 51 per cent of men and 41 per cent of women in 1974 to around 38 and 32 per cent respectively), and tobacco control was seen as a vote loser by politicians (Baggott, 1988, http://zen.lib.strath.ac.uk/images/strathimages/spogap/pdfs/057.pdf; ASH, 2005a, http://www.ash.org.uk/). This underpins Read's (1992) identification of a producer-dominated policy network. The practical basis for this dominance was the long-term relationship between the industry and government, cemented during World War II when TAC's predecessors were set up to ensure the supply of cigarettes. The TAC represented the four main domestic tobacco companies and could claim a high level of representational legitimacy. Direct TAC lobbying to senior ministers was supplemented by support from sports and arts ministers (due to the level of sponsorship) as well as MPs acting either as tobacco consultants or from constituencies with a strong tobacco presence (Taylor, 1984, 69–71). The TAC also enjoyed close relationships with senior civil servants in the Treasury and the Department of Trade and Industry (DTI) as well as day-to-day links at a lower level of government (for example, when negotiating the details of voluntary agreements). It was joined by organizations representing tobacco workers (such as the Tobacco Workers' Union), retailers, and even consumers, such as FOREST (Freedom Organization for the Right to Enjoy Smoking Tobacco). This gave the tobacco lobby ‘the appearance of a broad constituency of support’, with the TAC as its main voice to minimize the appearance of fragmentation (Baggott, 1988, 18). Newspaper advertising revenue, amounting to £17 million in 1981, the equivalent of £39 million in 2006 (Taylor, 1984, 49), ensured that few anti-smoking stories were published, and the TAC also had a close relationship with the (self-regulating) Advertising Authority, which feared that a ban would set a costly precedent (Baggott, 1988, 21).

Read (1992) suggests that this policy monopoly was surrounded by an issue network. In the 1950s and 1960s this included the Royal College of Physicians, followed in the 1970s by ASH and the British Medical Association (BMA). While the tobacco industry often engaged in debates within the wider issue network (calling on FOREST, the retailers, or unions), public health interests were excluded from the core network. Compared to the present day, ASH was not well funded, the BMA was not well organized, and cancer charities were not active (Baggott, 1988, 15). The anti-tobacco campaign was still in its infancy by the 1970s, and ASH could not be too critical of a government providing 95 per cent of its income (Taylor, 1984, 43). As a result, tobacco companies controlled the policy image of tobacco. The framing of tobacco as an economic issue — jobs, tax revenue, and exports — was the basis for support from the Treasury, the DTI, and the Department of Employment. Further, framing the issue as a matter of individual choice explained why increasingly accepted links between smoking and illness did not result in significant policy change.

Measuring Policy Change

Recent tobacco legislation therefore appears to mark a symbolic and substantive shift from the past, but how do we demonstrate or qualify this? One way is to situate these new regulations within a wider discussion of tobacco policy instruments and to use these categories to examine policy change over an extended period. This exercise suggests that accounts of change are intertwined with normative issues (how much change there should be) as well as competing narratives on the role and intentions of government. Adapting and extending Studlar (2004, 191) gives us eight relevant policy instruments:

  1. 1

    Regulation — on advertising, smoking in public places, sales of cigarettes, and levels of tar.

  2. 2

    Economic incentives (subsidies to farmers, tax expenditure on arts sponsorship) and penalties (taxation, litigation).

  3. 3

    Public education — including the ratio of health education to tobacco advertising.

  4. 4

    Smoking cessation services and nicotine replacement therapy.

  5. 5

    Funding external organizations (such as ASH).

  6. 6

    Funding scientific research (for example, through the Medical Research Council).

  7. 7

    Tackling tobacco smuggling.

  8. 8

    Levels of enforcement and the scale of punishment (particularly relevant to the history of voluntary agreements).

The point of identifying a range of policy instruments is that a focus on regulation alone exaggerates the extent of policy change. Looking at other measures also helps us to characterize the trajectory of change. The argument (below) that recent legislation marks an incremental step rather than a sea change in tobacco policy, for example, draws on the evidence of complementary policy measures.

Moreover, interviews with representatives from the main organizations involved in UK tobacco policy suggest a fundamental disagreement on what recent events represent — do they demonstrate successful pressure on a government reluctant to legislate and challenge tobacco interests, or do they reflect a logical progression from incremental steps already taken by government? Table 1 highlights this scope for different explanations by comparing two narratives of post-war policy change. While the ‘dominance narrative’ points to a closed policy community ensuring policy stability (even following New Labour's election in 1997), the ‘incremental narrative’ suggests a relatively open network which was more conducive to policy change.

Table 1 Contrasting narratives of tobacco policy change
Full size table

The Incremental Narrative

Three main points support a narrative of incremental policy change. First, there is evidence of a gradual move towards restrictive smoking policies from the early 20th century to the 1970s. Bans on the sale of tobacco to children (under 16) were introduced in 1908 and extended in 1933; the first smoking withdrawal clinic was established in 1958; cigarette advertising on TV was banned in 1965; the first voluntary agreement with the tobacco industry on advertising and health warnings on packs were introduced in 1971; and smoking was phased out in public transport and cinemas from the 1970s (ASH, 2005c, http://www.ash.org.uk/html/schools/keydates.html). Tax increases on tobacco justified on health grounds were also used extensively in the 1970s (Read, 1996, 70), and the nature of health education shifted from ‘stating the facts’ and letting individuals decide in the 1950s, to harm reduction (promoting ‘low tar’ cigarettes) in the 1960s, and to absolutism (there is no safe level of tobacco smoking) in the 1970s. Spending on health education concerning tobacco increased from £414,000 in 1973 (the cash equivalent of £3 million in 2006) to £2 million (£4.6 million) in 1981 (Taylor, 1984; Berridge and Loughlin, 2005, 960).

Second, Baggott (1988, 44–45) highlights a ‘hardening of the official stance’ in the 1980s, with more restrictive controls on voluntary advertising, stronger government health warnings, a bill tightening the law on sales of tobacco to children,Footnote 1 and taxation to discourage smoking in the 1984 and 1986 budgets. Voluntary agreements indicated ‘a change in the relationship between government departments and the tobacco industry’, providing benchmarks to assess compliance and signalling the inevitability of legislation (Read, 1996, 36, 54). Third, the election of a Labour government in 1997 marked a further advance. Tobacco control became a key part of its public health campaign following the publication of Smoking Kills (Cm 4177, 1998). This promised to implement the EU directive on tobacco advertising, increase spending on health education and smoking cessation services, combine increased taxation on tobacco with better customs controls, toughen the code of practice on smoking at work, and address smoking in public places with the voluntary Public Places Charter.

A review of tobacco control policies in 30 European countries, by the European Network for Smoking Prevention, suggests that the UK already has the second best record on the issue (Joosens and Raw, 2006). Further restrictions will elevate the UK to the top. The White Paper, Choosing Health, has already proposed to do this by legislating to phase out smoking in the workplace by 2007, with the exception of private clubs and bars not serving food (Cm 6374, 2004, 97–99, http://www.dh.gov.uk/assetRoot/04/09/47/60/04094760.pdf). On this basis, the government's argument is five-fold. First, that levels of tax on tobacco have peaked, given the threat of smuggling. Second, that policy could not get ahead of the scientific evidence, which did not set the effects of passive smoking in stone until the Scientific Committee on Tobacco and Health (SCOTH, http://www.archive.official-documents.co.uk/document/doh/tobacco/contents.htm) report in 1998. Third, that the approach has been deliberately incremental, with measures to influence, but not get ahead of, public opinion. The voluntary approach is tried, evaluated, and then replaced by legislation if it is not working. The fourth argument is that the voluntary approach was working in many areas. By 2004, 50 per cent of workplaces were smoke-free, 36 per cent had designated smoking areas, and 10 per cent involved people working outdoors. Excellent progress was also made on meeting smoke-free targets in restaurants. Finally, until recently, public health groups agreed with the need for incremental change. Therefore, the recent comprehensive ban is a surprising addition to, but logical progression of, existing government policy (interviews, Department of Health, 2006).

This incremental narrative suggests that the policy network was never as closed as Read (1992) suggests. Scientific and medical advice was always present, and ASH was a government-funded pressure group whose role was fostered by civil servants. Its exclusion from formal negotiations with government was caused by its unwillingness to work with tobacco companies (Berridge, 1999, 118). Further, what looked like government inactivity due to industry dominance was often a failure to accept epidemiological evidence or an unwillingness to risk the electoral fallout of legislation (ibid., 119). Nevertheless, the public health stance was eventually successful in changing demand for tobacco and attitudes towards policy. The constant production of evidence on illness, the demands of ASH, and the introduction of MP bills were used by government as leverage in negotiations with the industry. ASH's lack of formal contact also contrasted with its work behind the scenes with the Treasury (Read, 1996, 120), while DTI attitudes varied by minister (Berridge, 2004, 119).

The Dominance Narrative

The alternative ‘dominance’ narrative characterizes UK tobacco policy as a series of minimal responses to public health pressure. From 1956 to 1959, the Ministry of Health spent less than £5,000 (£70,000 in 2006 terms) on anti-smoking messages compared to the £27 million (£376 million) spent on tobacco advertising (Taylor, 1984, 5). The demands from government on the industry were light, with the use of filter tips in cigarettes and funding to the Medical Research Council (£250,000, or £3.5 million in 2006) being the most significant (Read, 1996, 43). Although epidemiological evidence of the link between smoking and illness was accepted by the government in the 1960s, any policy change still required chief medical officers and health ministers working together. So, while George Godber (the Chief Medical Officer from 1960 to 1972) encouraged the Royal College of Physicians to publish reports on the links between smoking and health, the results were not seen by the Health Minister, Enoch Powell, as a sufficient reason to legislate in order to restrict advertising. The ‘worst blows’ to the tobacco industry in the 1960s were health warnings on cigarette packs and a ban on TV advertising (Taylor, 1984, 7). Yet the former helped the industry refute legal claims based on ignorance of the health effects of smoking, while the latter showed the limits to ministerial action. The ban was possible only because a new Health Minister, Kenneth Robinson, found an ally in Tony Benn (the Postmaster General) who could act without the need for legislation by directing ITV's advertising watchdog (ibid., 82). The industry responded with a gift coupon scheme (which reversed the falling consumption of cigarettes) and sports sponsorship (with no health warnings) to circumvent the ban and extend the potential to advertise to children (thus undermining other agreements). More extensive legislation was never achieved. Richard Crossman, as the first Secretary of State for the new Department of Health and Social Security in 1968, overruled Robinson by pushing for voluntary agreements over legislation. These were chosen not because they worked but because legislation would be unpopular. Since the threat to legislate had no political weight, there was little incentive for tobacco companies to take the agreements seriously (Taylor, 1984).

The most significant progress in the 1970s, relating to smoking on public transport, was caused by concerns for safety rather than health (Read, 1996, 7). The threat of legislation was not revisited until 1974 by the Health Minister, Dr David Owen. His plan was to classify tobacco as a medicine and therefore control its content and promotion. Yet the lack of a working majority in Westminster hindered legislative progress, and Owen had become Foreign Secretary by the time a majority was achieved (Taylor, 1984, 94). Within the post-1979 Thatcher government, Sir George Young was an active but junior health minister who moved for a ban on advertising and sports sponsorship. This was seen by many Conservative MPs and ministers (including Thatcher herself) as an attack on ‘freedom not cigarettes’. The proposal was rejected and Young was shunted to another department (ibid., 145).

The history of voluntary agreements is therefore one of slow movement and limited government ‘bite’. Many policy changes were easy for the industry to accept — introducing filter tips, which helped marketing to women and persuaded some that they were smoking safely; stopping advertising on high-tar cigarettes, which had a low and declining market share; and limiting advertising expenditure, which suited the companies with the highest market shares. Other measures, such as health warnings on cigarette packs, were traded for the ability to use brand names in advertising (Read, 1996, 58). Moreover, while advertising expenditure was capped, this was neither policed effectively nor matched by health education spending. Most Health Ministers used the agreements to create the impression that they were taking action, with the Treasury on hand to block those who tried to go further (ibid., 50–54).

Post-war UK policy contrasts with greater advances in the US and Norway (Baggott, 1988; Read, 2005). Successive measures were incremental but were not characterized by Lindblom's ‘partisan mutual adjustment’, since there was dominance by one set of interests and the starting point did not represent a negotiated balance between industry and health. The use of voluntary agreements was a victory for the tobacco lobby since they ensured its continued access to government and control over the level of implementation. After the election of New Labour in 1997, the last voluntary agreement took 10 years to replace. In 1999 Labour introduced a new voluntary code and: ‘Health bodies were up in arms. All pubs had to do was put a sticker up in the window to say it was a smoking pub. So everyone put up stickers!’ (interview, ASH Scotland, 2004).Footnote 2 A very small proportion of pubs went smoke-free, while those with non-smoking areas relied on ventilation systems that the tobacco industry knew were limited (Leavell et al., 2006, 227–228).

Narratives and Models of Policy Change

These alternative narratives show the level of competition to define the nature of policy change. Since both present convincing accounts of policy development, we need to be careful about assuming a policy's history before examining recent events. This point is crucial to the explanatory power of models of policy change: their value is inextricably linked to the narrative of policy change that we select. We can demonstrate this in a discussion of two main types of explanation. The first directs attention to influence from ‘above and below’ the UK. This is most valuable if we select the dominance narrative that emphasizes post-war policy stability and inertia. According to this model, recent change has been profound and has only been made possible by a successful challenge to existing relationships, helped by events external to the network (such as the role of the EU). Explanations highlighting external influence are less impressive, however, if we select the incremental narrative which suggests that recent legislation is a logical progression from existing policy with an established trajectory. This maintains that external influences had little effect on policy change. In contrast to ‘above and below’ accounts, the second type of explanation uses policy networks analysis to account for periods of stability and change. While the Advocacy Coalition Framework (ACF) can be consistent with both narratives, punctuated equilibrium depends on the dominance narrative of policy development.

Policy Change from Above and Below

UK tobacco policy follows the broad trends identified in Studlar's (2004) analysis of developed countries. Most have advertising bans on TV and radio (with sponsorship more difficult to address), cigarette warning labels, health education campaigns since the 1960s, and higher taxation levels since the 1980s. Full advertising bans and a prohibition on smoking in public places have only recently become realistic issues. A common theme, too, is the post-war debate on the links between smoking and illness, replaced by more established medical evidence, with relentless challenges by the tobacco industry to the policy implications (Feldman and Bayer, 2004, 1). Responses have varied according to the ‘vested economic interests, cultural practices, and political factors’ of each country (Studlar, 2004, 215). Therefore, we need to dig deeper to find the causal mechanisms involved. Models stressing influence from above-and-below include:

  1. 1

    Venue shift, in which binding decisions made in venue B affect policy in venue A (Baumgartner and Jones, 1993, 32).

  2. 2

    Multi-level governance (MLG), which describes the dispersal of power from central government to other levels of government and non-governmental actors (Bache and Flinders, 2004).

  3. 3

    Policy transfer and learning.

While these models help us to understand the links between decisions made elsewhere and their effects on UK policy, the extent of this influence varies. This is categorized by Dolowitz and Marsh's (2000) discussion of policy transfer, which involves:

  • Direct coercive transfer, which can involve a supra-national body (such as the EU) taking over responsibility for policy development and obliging individual countries to follow.

  • Indirect coercive transfer, which describes a perception within region A that it should follow the policy of region B.

  • Voluntary transfer, which describes the relative freedom to interpret and learn from decisions made elsewhere.

Elements of all three are apparent with the EU's influence on UK tobacco policy (although this has yet to extend to public places). This began with the European Commission's Europe Against Cancer campaign and directives banning tobacco advertising on TV, tobacco product labelling, limits on the level of tar in cigarettes, and minimum tax levels. The most ambitious initiative was the first Tobacco Advertising Directive (TAD1), which banned advertising in the print media, radio, and internet as well as in cinemas, posters, and merchandise-based advertising. This passed with a narrow majority in 1997, but Germany (backed by the tobacco industry) referred the issue to the European Court of Justice (ECJ) in 2000 (Duina and Kurzer, 2004, 58). The ECJ repealed the directive but left the door open for a more limited directive covering advertising with a cross-border element (TAD2, adopted in 2002).

The effect of this directive on the UK is not straightforward. There is coercive transfer to an extent, with a proxy counterfactual provided by Germany which will now implement TAD2 following an unsuccessful appeal. However, the Labour party was already committed to banning tobacco advertising in its 1992 manifesto.Footnote 3 Kevin Barron MP attempted to do so in a Private Member's Bill in 1993. Although this fell, its measures were adopted as Labour policy and included in the 1997 manifesto, with Barron appointed as the Shadow Public Health Minister. Further, the 2002 Tobacco Advertising and Promotion Act implements most of the requirements of TAD1 without being required to do so. Yet, Labour's actions in government differed to its public statements in opposition. Even before the 1997 election, Labour was nervous about the issue of sports sponsorship, with shadow ministers pursuing a form of words to ‘get them off the hook’ (interview, Labour MP, 2006). When in government, Barron (who maintained a relatively hard line on sponsorship) was passed over for the ministerial post. The UK then voted against TAD2, with some suggestion that this related to a £1 million donation to the Labour Party by Formula 1's Bernie Ecclestone (Duina and Kurzer, 2004, 70). The UK government was in no hurry to legislate and the bill followed a convoluted path, proposed in 2000 but running out of time before the 2001 dissolution of Parliament (Read, 2005).

This provoked a response from ‘below’, with the new legislative powers granted to Scotland demonstrating the potential for venue shift as a means of exerting pressure on UK policy. Originally the UK bill was to cover Scotland, but when it fell Nicola Sturgeon MSP introduced a separate (more limited) Scottish Member's Bill. This was debated in the Scottish Parliament throughout 2002, but then withdrawn in favour of a single UK Act. Given the uncertainty over Scottish competence in this area, the Scottish bill was used primarily for agenda setting, with MSPs lobbying their MP counterparts on the issue (Cairney, forthcoming, 2007). Stronger Scottish influence is apparent with smoking in public places. In 2005, Scottish Executive legislation marked a clear departure from the UK's voluntary approach. This was subsequently used by public health groups in the UK as leverage for change (interviews with ASH, BMA, CIEH, 2006). Much was made of England appearing to fall behind Scotland (with Wales and Northern Ireland signalling similar intentions), particularly since Scotland's implementation date coincided with the vote in Westminster. Therefore, in both examples we see the potential for influential venue shift afforded by multi-level governance arrangements.

But what is the upshot of this influence? The dominance narrative suggests that with advertising we see a significant impetus to act from above and below, and with smoking in public places we see pressure based on embarrassment. Yet, the UK government introduced legislation on advertising beyond the minimum requirements of the EU. Scotland was not the only source of pressure, and interviews with MPs suggest that Sturgeon's bill barely registered in Westminster. Pressure from the House of Commons and the Lords, combined with the embarrassment of the Ecclestone affair, prompted the government to act. Similarly, the UK government had chosen not to follow Scotland with a comprehensive ban on smoking in public places and this point was made forcefully by the Health Secretary, John Reid, to Scotland's First Minister, Jack McConnell (The Scotsman, 29 September 2004). The Department of Health view (interview, 2006) was that international policy change was incremental. Bars and clubs were left until last, but were clearly isolated until public opinion was conducive to this final change. A quicker path chosen by Ireland and followed by Scotland was not the accepted model. Reid's actions demonstrated that the UK government could voluntarily choose from two policy-learning options. It chose to follow the international experience closest to its existing (incremental) policy. Therefore, external venue shift alone does not fully explain policy change on smoking in public places.

Policy Networks and External Factors

A key concern of policy networks analysis is the relationship between ‘internal’ explanations for policy stability and ‘external’ reasons for policy change (Marsh and Rhodes, 1992, 260–261). External factors may include:

  1. 1

    Ideological change following the election of a new government.

  2. 2

    Change from ‘above and below’.

  3. 3

    Changing information, including medical evidence and the experience of international policy change.

  4. 4

    The changing economic benefits of tobacco:

    • The number of jobs in the UK directly related to tobacco fell from 40,000 in 1979 to 11,000 in 2003 (ASH, 2005a).

    • Tobacco tax in 1996 was one-quarter of the value in 1950 as a proportion of total revenue (Berridge, 2004, 130).

    • Rising imported and illegally imported market shares. Before the 1980s almost all tobacco consumed in the UK was from a domestic source, with UK tobacco consumption supporting UK employment (Baggott, 1988, 45).

  5. 5

    Social change:

    • The drop in smoking prevalence from 51 per cent of men and 41 per cent of women in 1974 to 28 per cent of men and 24 per cent of women in 2005 (ASH, 2005a).

    • Changing public attitudes: those in favour of smoking restrictions in pubs rose from 48 per cent in 1996 to 65 per cent in 2004.Footnote 4

The key for policy networks analysis is to show that these factors influence the direction of policy but do not determine it. The common element is mediation — the weight of interpretations placed on these factors by decision makers and pressure participants.

The Advocacy Coalition Framework

The ACF focuses on sectoral level subsystems that include more actors than policy communities: ‘not only interest group leaders, but also agency officials, legislators from multiple levels of government, applied researchers, and perhaps even a few journalists’ (Sabatier and Jenkins-Smith, 1993; Sabatier, 1998, 103). The glue that binds actors within competing coalitions is ‘belief systems’. These range from ‘core’ beliefs (such as the relative priorities of freedom and health), ‘policy core’ (the proper scope of government), and secondary aspects (the best way to deliver policy). Core values are the least susceptible to change — ‘akin to a religious conversion’ (1993, 221) — while policy beliefs may only change following external ‘shocks’ to the subsystem (such as changing socio-economic conditions). Secondary aspects are more subject to change following policy learning (such as environmental policy shifting from command-and-control to economic incentives). These beliefs are refined according to new information and the ‘enlightenment function’ of policy analysts. Advocacy coalitions not only compete for position within subsystems (with a role for a neutral ‘policy broker’) but also revise their strategic positions based on new evidence and the need to react to external events. We therefore have stable and dynamic elements. Stability comes from the parameters of policy — the constitutional structure, fundamental social structures and values — and perhaps from dominance by one coalition. Change comes from reactions to external events (which may undermine dominance) and the assimilation of new evidence (mediated by existing beliefs).

The ACF replaces the idea of a producer network within an issue network. Rather, we have pro- and anti-tobacco coalitions. While pro-tobacco dominated the post-war period, it still engaged with anti-tobacco in competition to interpret information and seek favour from the policy broker. There was no ‘partisan mutual adjustment’ in the early post-war period. Rather, adjustment (or policy learning) is made by a dominant coalition in the face of changing information and external environments. More significant policy change comes from external shocks to the system — a new government with different ideas, increasing EU influence, or shifting public opinion. These shocks are mediated, with the pro-tobacco coalition adapting (or learning) to maintain its dominant position — introducing filters for cigarettes, funding medical research, voluntarily restricting advertising, and providing ventilation in public places. So, while the anti-tobacco coalition may now dominate the subsystem, the value of ACF is in explaining why such a shift took so long. Much of the delay was achieved through the constant re-appraisal of new evidence, from the post-war rejection of the scientific evidence on illness, to more recent scepticism about the level of risk from passive smoking and what constitutes a proportionate response.

However, there are problems with this interpretation. First, the ACF explains stability better than change. Policy change of any magnitude tends to be explained by external shocks rather than the coalitions themselves (Cairney, 1997, http://www.psa.ac.uk/journals/pdf/5/1997/cair.pdf; John, 1998). Second, the framework struggles to explain temporary alliances based on self-interest rather than core beliefs. For example, the licensed trade was pro-tobacco until 2005 since a smoking ban would harm business. When the government proposed to ban smoking in pubs but exempt ‘private clubs’, the British Beer and Pub Association joined the lobby for a comprehensive ban. Its interests were served at different times by different coalitions. Similarly, while civil liberty groups may join the pro-tobacco coalition on the basis of core beliefs (a limited role for the state), there is no demonstration that tobacco companies are driven by these beliefs. Indeed, a traditional criticism by domestic tobacco was that government action was not extensive enough, for instance, on import duty and customs. The situation of government departments is more problematic. We have two possibilities: first, the Treasury was pro-tobacco and the Department of Health anti-tobacco. Yet the Treasury's driver for rejecting tobacco regulation in the past (lost revenue) does not fit in with the policy core beliefs of the pro-tobacco coalition any more than the tobacco companies’. The second possibility is that government departments perform the policy broker role. This fits with the Department of Health's role in negotiating voluntary agreements. Yet, health ministers were often constrained by more powerful departments and excluded from the core network (Read, 1992). This suggests that policy brokerage — a benign concept within the ACF — is more important to the success of coalitions than their own strategies. Finally, since the coalitions are broad it is difficult to track the significance of venue shift. Within the ACF the constitutional structure represents stability. However, the evidence from multi-level governance is that the constraints provided by constitutional structures are fluid, providing more change than stability.

Punctuated Equilibrium

The focus of the punctuated equilibrium approach is to explain long periods of policy stability punctuated by short but intense periods of change. Baumgartner and Jones (1993) suggest that since decision makers, the media and the public all have limited resources (time, knowledge, attention) they cannot deal with the full range of ideas or policy problems. So they ignore most and promote a few to the top of their agenda. Resource constraints also limit attention to certain aspects of a policy problem. Problem definition is crucial since it determines the level of attention and the nature of government response. This process explains tobacco policy monopolies and the ability to challenge them. Tobacco companies pursue the definition of a policy problem (or its policy image) as a boost to the economy. This limits the number of participants who can claim a legitimate role in supporting policy. Those excluded from monopolies have an interest in challenging this image. The role of new ideas (such as passive smoking) or new evidence (such as the experience of Ireland) is crucial to divert attention to other aspects of the same problem. For example, if the scientific evidence associates smoking with ill health and attention shifts to minimizing harm, the decision-making process widens to accommodate new experts. If this new image is stifled by policy monopolies, then groups pursue ‘aggressive venue shopping’ to seek influential audiences elsewhere, such as the courts, other types of government, the media, and the public (Jones and Baumgartner, 2005, 5).

Baumgartner and Jones (1993, 93) describe this process in the US. In the early 20th century, tobacco attracted minimal media attention and most government attention was favourable. Tobacco enjoyed a glamorous image and consumption was high. However, since the 1960s, we have seen heightened and negative media coverage and a drop in cigarette consumption. This negative attention causes a reappraisal of the positive aspects — for example, the economic benefits are undermined by a focus on rising health insurance and decreasing worker productivity (ibid., 114). Similar shifts of media attention, smoking prevalence, and public attitudes are apparent in the UK, reflecting increased acceptance of the scientific evidence linking smoking (and then passive smoking) to illness. Baumgartner and Jones (1993, 87; Jones and Baumgartner, 2005) suggest that there is a direct causal link between this type of attention and rapid policy change. Peak periods of organizational change ‘generally coincided with Gallup Poll data showing public concern with the same problems’.

However, this demonstration of causality is problematic. First, while the US displayed high and negative levels of attention and possessed the most organized public health lobby, the federal response to tobacco has rarely been as intense. Before the 1990s, most policy progress was achieved through the courts or devolved levels of government (Studlar, 2002). In the UK, fewer influential venues were apparent before the late 1990s and even now their influence is uncertain, particularly since devolved policy differences are more constrained within a unitary state (Cairney, 2006a). Second, there is no demonstration that public and media attention determines the nature or intensity of governmental response. A discussion of narratives reinforces this point. The dominance narrative suggests that the post-war policy response was to minimize wider demands for change. Public health concerns (reinforced by public and media attention) were addressed with a combination of voluntary agreements and an appeal to individual choice. While policy appeared to change, the key policy instruments were never enforced. The 20th-century policy development is characterized by unsuccessful public health attempts to shift the policy image of tobacco within government, with little recourse to alternative venues.

If we follow the incremental narrative we see more evidence (from the 1960s) of a changing policy image within government, as attention shifted from the economic benefits of tobacco to the scientific evidence on illness. This was increasingly accepted by health ministers who took steps to limit the acceptability and prevalence of smoking. However, this was not caused by venue shift (since Parliament was often more sympathetic to tobacco interests and influence from the courts was non-existent) or by public and media opinion at the time (since these steps were taken in spite of the electoral consequences). A more convincing explanation is that the network was never closed to health interests (particularly since the BMA was the Ministry of Health's main ‘client’) and adaptation to new evidence was based on the British policy style of ‘bureaucratic accommodation’ (see below).

The dominance narrative is more supportive of a recent punctuation regarding the legislation on smoking in public places. Internal stability was initially maintained by ineffective voluntary agreements, even following the election of a Labour government in 1997 and a definitive statement on passive smoking by SCOTH in 1998. Then, devolution and the increased scope for ‘venue shopping’ led to the prospect of comprehensive legislation in all UK countries except England. This contributed to increasing levels of public attention to passive smoking and shifted the policy image of tobacco (including the balance of opinion between freedom and public health). Almost all of the interviews conducted for this study point to rapid public opinion change (following the experience of Ireland and Scotland) as a key factor in policy change. The nature of attention limited the governmental response since the appeal to individual freedom was no longer consistent with the new policy image.

Yet, the incremental narrative qualifies the significance of these external influences. As the discussion of ‘above and below’ suggests, while the Department of Health was already committed to policy change, it rejected the approach taken by Ireland and the rest of the UK. A degree of mediation is also apparent with public opinion. Curtice (2006, 57) suggests that the Scottish Executive went ahead of public opinion with comprehensive legislation. This was helped by the results of a huge consultation giving legitimacy to a complete ban (Cairney, forthcoming, 2007). In contrast, the UK government followed public opinion to the letter:

Surveys…show 86% of people in favour of workplace restrictions, and a similarly substantial majority of people supporting restrictions in restaurants. But when people are asked whether smoking should be restricted in pubs the figures fall substantially — to around 56% — and when people are asked which sort of restrictions they would prefer in pubs only 20% of people choose ‘no smoking allowed anywhere’ and the majority tend to be opposed to a complete ban (Cm 6374, 2004, 98).

This line did not change when the vast majority of respondents to public consultation called for a comprehensive ban (Department of Health, 2005, 2006, http://www.dh.gov.uk/assetRoot/04/12/36/21/04123621.pdf, http://www.dh.gov.uk/Consultations/LiveConsultations/LiveConsultationsArticle/fs/en?CONTENT_ID=4136732&chk=91z6mN) or when subsequent surveys revealed a hardening of public opinion on smoking in pubs. Therefore, external pressure alone does not explain the extent of the ban since there is enough scope for governments to choose the information they base decisions on. A comprehensive ban was resisted by John Reid when Health Secretary. When Reid became Secretary of State for Defence, he was still instrumental in the Cabinet decision to reject a compromise (only exempting private clubs) proposed by his successor, Patricia Hewitt. Therefore, a full explanation of comprehensive legislation requires analysis of venue shift to Westminster. Further, since (according to policy networks analysis) the scope for parliamentary influence tends to be limited, we also need to explain the particular circumstances which led to the centrality of Parliament in this case.

Multiple Streams

Kingdon's analysis adds an extra factor to the explanation. We have the redefined policy problem (tobacco as a pressing public health issue) and a solution (an advertising ban and a comprehensive ban on smoking in public places), but we do not yet have the explanation for the adoption of that solution. As Kingdon (1984, 165–6) argues, policy change results from the synthesis of all three:

Separate streams come together at critical times. A problem is recognised, a solution is developed…a political change makes it the right time for policy change, and potential constraints are not severe…these policy windows, the opportunities for action…present themselves and stay open for only short periods.

This analysis highlights the detailed, idiosyncratic reasons for policy change. But what factors opened these windows? In advertising the process is relatively clear, with the introduction of an EU directive and a Labour government committed to implementing it. The window eventually opened following Scottish influence, the initiation of the bill in the Lords, and parliamentary pressure fuelled by the Ecclestone affair. However, the factors ensuring a comprehensive ban on smoking in public places require more discussion. First, if we follow the dominance narrative (highlighting the marginalization of health ministers), then we see a shift in Department of Health influence when John Reid was appointed in June 2003. Reid was a strong Secretary of State (supported by Prime Minister Tony Blair) who made public health a priority and who ensured that any decision on smoking would be made by the Department of Health. This gave greater prominence to its Chief Medical Officer who in the past had been crowded out in tobacco discussions. Liam Donaldson (the CMO since 1998) was particularly active in this area, highlighting the issue in annual reports, criticizing UK policy in comparison with other countries, criticizing the socio-economic effects of a partial ban, and reporting that he came close to resigning following the Cabinet's reluctance to opt for a complete ban.Footnote 5

The centrality of the Department of Health accelerated the shifting resources of pressure participants. There were already signs of the declining influence of tobacco interests given the reduction of employment and revenue from tobacco. Further, the Treasury had become concerned with the broader productivity and health costs of tobacco consumption. This is apparent in the Treasury's health inequalities policy as part of its coordination of prevention strategies across Whitehall departments. It identified smoking as ‘the single most significant causal factor for the socio-economic differences in the incidence of cancer and heart disease’ (HM Treasury and Department of Health, 2002, http://www.hm-treasury.gov.uk/media/1F8/DC/Exec%20sum-Tackling%20Health.pdf). The old argument about anti-tobacco policies victimizing the working classes was turned on its head. Reducing smoking among ‘manual social groups’ was highlighted by the Treasury as a key means to reduce differences in life expectancy. Therefore the nature of Treasury influence has changed and is less likely to undermine tobacco control. The influence of tobacco companies (and the licensed trade) through the DTI was also curtailed following Reid's appointment. In contrast, the presence and resources of public health groups have strengthened. The BMA and ASH have a stronger campaigning alliance and were joined by other organizations.Footnote 6 The role of Cancer Research UK is particularly significant since, while this issue marks its first attempt to ‘lobby’ government, it has a strong reputation on research and extensive links to the Department of Health, the Treasury, and MPs (interview, CRUK, 2006). Further, the Chartered Institute of Environmental Health argued that it would be costly to enforce a partial smoking ban (interview, 2006), while the focus on health and safety ensured a key role for the Trades Union Congress.

The background to the second factor crucial for enabling reform, namely parliamentary influence, was a series of ‘rebellions’ by Labour MPs on issues such as education reform and ID cards. The prospect for further revolt (and a Cabinet split) on an issue not high on the government's agenda was key to the decision to allow a free vote (which was crucial given Conservative opposition to legislation). The threat of revolt was furthered by Kevin Barron, the chair of the Health Select Committee. Barron was instrumental in undermining Reid's insistence that Labour MPs stick to the 2005 manifesto commitments (by publicizing evidence suggesting the manifesto did not derive from Labour's 2004 National Policy Forum). He then secured committee time to examine the legislation and highlight issues to MPs (including Donaldson's threat to resign). He secured an agreed line from the committee report and was able to ‘deliver’ a large number of MPs willing to vote for a complete ban (interview, Barron, 2006). Barron and David Taylor (the chair of the All-Party Parliamentary Group on Smoking and Health) then met Tony Blair in November 2005 and persuaded him that a free vote would be popular and would rescue some leadership credibility within the party (interview, Taylor, 2006).

The level of Labour MP support was achieved following an unusual amount of pressure participation. The BMA and ASH targeted MPs directly and through the local media, while local doctors applied pressure at the constituency level. The campaign dominated the lobbying time for these groups, all coordinating their efforts towards a simple health and safety argument — that if an MP accepted the argument on the health of workers in restaurants, it could not ignore the health of workers in bars and clubs. This was bolstered by the success in Ireland (Barron and Taylor had organized MP visits) and the prospect of full bans in the rest of the UK. The result was that an overwhelming majority voted against John Reid's exemptions and for comprehensive legislation (Cowley and Stuart, 2006, http://www.revolts.co.uk/Smoking%20Vote%2014%20Feb%2006.pdf).

Are These Results Generalizable?

The reliance on idiosyncratic reasons for policy change leads to the problem of generalizability. There are a number of reasons to suggest that these findings are not reflected widely in British politics. First, the level of MP interest on this particular issue is unusual, and we should be cautious about making broad conclusions about parliamentary influence during an extended period of Labour rebellion (Cowley, 2006, 55). Second, the nature of pressure politics was also unusual.Footnote 7 Tobacco took up a disproportionate amount of BMA time and marked a departure in strategy by cancer charities (interviews, BMA and CRUK, 2006). The case study also highlights the success of an ‘open strategy’, or maintaining multiple channels of access to government, parliament and the public (Whitely and Winyard, 1987, 86–87). This contrasts with the policy communities literature stressing insulated contacts between government and groups who agree to ‘sell’ the results of negotiations regardless of the level of disagreement (Jordan and Richardson, 1987; Marsh and Rhodes, 1992). The traditional ‘British policy style’ was to seek consensus through ‘bureaucratic accommodation’ removed from the glare of public attention. This process often took place between conflicting groups following the ‘realization that both sides could “win”’ (Jordan and Richardson, 1981, 80–81; Jordan and Maloney, 1997, 578; Cairney, 2006b). ASH was successful despite flouting these rules, but in this case the Department of Health was sympathetic and ASH's role was complementary (tending to criticize the effects of tobacco and the industry rather than the government). In cases with clearer opposition to government policy, the results are different. Groups in the voluntary sector know that criticism affects their funding and status within the network; education unions are only included in negotiations on workforce issues if they sign the ‘social partnership’ prohibiting public criticism; and an open strategy against the Mental Health bill led to the loss of legislation (in 2006) rather than a change in policy. Third, policy transfer in England based on pressure from the rest of the UK is unusual. In most cases (health, housing, and higher education in particular), the pressure flows in the opposite direction. Fourth, the influence of the EU varies, from direct and strong in agricultural and environmental policy, to indirect and variable in health and education. Therefore, the value of each model of policy change is likely to vary by case study. This reinforces the value of a multiple lenses approach.

Conclusion

UK tobacco policy appears to be characterized by rapid change following long periods of stability. To an extent we can explain stability with reference to post-war socio-economic conditions: tobacco was an important source of jobs and revenue, smoking prevalence was high, and anti-smoking legislation was perceived to be unpopular. However, there is disagreement about the effect this had on policy. The dominance narrative suggests that a producer-dominated policy network was formed around the definition of tobacco as an economic issue. Public health interests were excluded from the core network. Although the evidence linking smoking to illness was increasingly accepted within government, the policy response was minimal. The definition of smoking as an issue of choice ensured that voluntary agreements (rarely enforced) were chosen over legislation. In contrast, however, the incremental narrative suggests that the exclusion of science and medicine was exaggerated. Policy change was incremental but on a clear path towards tobacco control. While the policy response to advertising and public places was often limited, it signalled that legislation would follow if self-regulation failed.

The narratives serve two crucial functions in the explanation of policy change. First, they promote clarity on the extent and meaning of change. The dominance narrative describes recent legislation as a sea change in policy and a challenge to vested interests and inertia in government. This is challenged by the incremental narrative which sees legislation as a logical progression of policy. Second, this discussion is crucial since the nature of change determines the value of explanatory models. Models that highlight the influence of external factors are most impressive if we identify long-term internal stability. If the UK was already on a clear path towards tobacco control, then their value is less clear.

The extent to which tobacco advertising legislation was caused by events ‘above and below’ varies by narrative. There is evidence of multi-level governance in the dispersal of power to the EU and Scotland, with venue shift producing policies that influenced the UK. The need to implement an EU Directive coupled with pressure from Scotland suggests that a recalcitrant UK government was coerced into action. However, since Scottish activity did not particularly register in Westminster and the UK's legislation went beyond EU requirements, there is a high degree of voluntary transfer consistent with the incremental approach. This argument is clearer on the issue of smoking in public places, on which there is little EU pressure. While the prospect of a comprehensive ban in the rest of the UK was embarrassing and the evidence from Ireland was impressive, the UK government drew on a wider international evidence base which was more of a fit with its approach.

Policy networks analysis suggests that while socio-economic shifts strengthen public health interests, policy change is not inevitable. With the ACF we see pro-tobacco weakened, but the long-term propensity of the policy broker to accept its core argument (economy and freedom over health) explains why change was limited. In the face of external shocks, pro-tobacco was willing to adapt to maintain its dominance, and has done so successfully for decades. However, the ACF struggles to explain significant change as well as stability. Change comes from external shocks but also constitutional pressures, which to the ACF are sources of stability. The lesson from multi-level governance is that the fluid constitutional position is as much a source of change, with venue shift often a key determinant of government policy.

Punctuated equilibrium explains stability with reference to a dominant policy image within a policy monopoly. Change results from a challenge to that image, often by appealing to venues outside of the policy monopoly, to widen participation and focus attention on different aspects of the policy problem. However, it is difficult to identify this punctuation in post-war UK tobacco. The dominance narrative suggests that policy change was minimal and there were few influential venues outside government. While the incremental narrative points to a shift of attention and problem definition within government, this was despite public opinion and the lack of alternative venues. Rather, the policy network was never closed to health interests and policy changed through the normal process of bureaucratic accommodation. The dominance narrative is more sympathetic to the idea of a more recent punctuation caused by rising attention to passive smoking. By the early 2000s the policy environment had changed, with devolution providing scope for external influence. Public attention to the effects of passive smoking (and policy in the rest of the UK) is cited by most interviewees as a key reason for policy change. This new image of tobacco was crucial since it limited the choices available to government. However, the incremental narrative points to the mediation of public pressure. The evidence suggests that while public opinion was changing quickly from 2004, the UK government line did not.

In all cases, a discussion of the unusual significance of venue shift to Westminster is necessary to fully explain the adoption of comprehensive legislation. This depended on a series of events that came together at the right time — rapid shifts of public opinion; high levels of pressure participant activity, a strong Health Secretary, and previous Labour rebellions ensuring a free vote. Without the latter, the Labour whip combined with Conservative opposition to legislation would have ensured the more limited government policy.

This reliance on idiosyncratic explanation may undermine generalizability and the results of this case study have qualified relevance to the study of UK public policy as a whole. However, variations by policy sector and sub-sector over time are a permanent feature of policy analysis. This variation affects the value of different models of policy change and therefore reinforces the significance of a multiple lenses approach.