Under Which Configurations of Streams Do States Learn? The Adoption of Tobacco Advertisement Bans in the Swiss Case as MSF Decision

Transcription

1 Under Which Configurations of Streams Do States Learn? The Adoption of Tobacco Advertisement Bans in the Swiss Case as MSF Decision Johanna Künzler 1 Paper prepared for the 24th World Congress of Political Science, Poznan, July 2016 Panel How do Policymakers Learn? Configurations of Political Learning and Policy Learning in Public Policy Analysis Abstract Advertisement bans for tobacco products are a well-established form of tobacco prevention policy. They constitute a widely acknowledged instrument within the respective expert communities and are actively promoted by the WHO all over the world. Despite this broad support however the bans frequently meet with difficulties when they are to be adopted in a political system. In Switzerland, tobacco ad bans were introduced in 15 out of 26 cantons after the canton of Geneva had won at the Federal Court with its attempt to introduce a respective law. The spread of the bans took place within seven years from this Federal Court decision and has stopped since then. Consequently, we consider the winning of the Geneva lawsuit a window of opportunity for cantonal promotors of ad bans to push it in their respective cantons. Adopting a corresponding MSF perspective, we use the fact that the Swiss cantons are fully fledged yet comparable political systems due to the common national context for the analysis of the interplay of political and policy factors that determine learning in political systems. It appears that learning is impeded by specific constellations of political factors. A comparative examination therefore poses a unique opportunity to grasp what it is exactly that promotes or hinders policy learning. We complement MSF with institutional and organisational factors and employ configurational QCA techniques to study constellations of streams that explain the adoption or nonadoption of tobacco ad bans in the Swiss cantons. The paper hence focuses on configurations of streams rather than mono-causal explanations for learning. 1 Johanna Künzler, University of Bern, KPM Center for Public Management, johanna.kuenzler@kpm.unibe.ch 1

2 Introduction Tobacco is the only legal product worldwide that kills half of its consumers, even when used according to the manufacturers recommendation (European Commission 2004: 89). Due to the 6 million deaths per year that are directly or indirectly attributable to the consumption of tobacco products, the World Health Organization (WHO) labels tobacco use a global epidemic" (World Health Organization 2013: 12). On 21 May 2003 the WHO s General Assembly unanimously adopted the Framework Convention on Tobacco Control (World Health Organization 2009: 18). 176 countries thus committed themselves to protecting present and future generations from the health, social, ecological and economic hazards emanating from tobacco consumption (World Health Organization 2003: 5). Besides other well-known measures such as the taxation of tobacco products or the regulation of ingredients, the agreement also recognizes the effectiveness of advertisement bans (World Health Organization 2003: 11f.). It has been shown that tobacco advertisement is a potent means of recruiting new consumers (World Health Organization 2013: 13; Mavrot et al. 2016: 2). Prohibiting it could therefore prevent at least a part of the population from engaging in addictive behaviour. The countries that have signed the Framework Convention agreed to prohibit tobacco advertisement or at least to limit and control it. In Switzerland one of the convention s signatories a quarter of all inhabitants above the age of 15 consider themselves to be smokers (Gmel et al. 2015: 46). In spite of this, no action has of yet been taken by the national government to introduce a law against tobacco advertisement 2. In a European ranking, this results in Switzerland being on the last position of 34 countries when it comes to the ban of tobacco advertisement (Joossens and Raw 2014: 17). In the absence of guidelines from the federal level, some of the subnational units have decided to act on this issue autonomously. The spread of laws against the promotion of tobacco products was triggered by a decision of the Swiss Federal Court in 2002, which explicitly stated that it was within the cantons legal competence to do so. By the end of 2015, 15 out of 26 cantons had introduced such a law 3. Interestingly, they are far from uniform: Some cantons prohibit billboard advertisement in the public area only, while others for example include the ban of spots in cinemas 2 In fact, 12 years after signing the WHO convention the Swiss government last November published a legislatory draft on the prohibition of tobacco advertisement. The final decision by the Parliament is still outstanding, however one of the two chambers in June already pronounced its refusal (Forster 2016). 3 accessed on

3 or the sponsoring of cultural events as well. Meanwhile, some of the jurisdictions never bothered discussing or adopting a law at all. The study at hand sets out to find an explanation to this wide variation in cantonal policy decisions. The issue of policy decision output is of vital importance to the field of policy analysis: Policy analysis is contingent on explaining policy outputs [ ], where the challenge is to account for different drivers and pathways that influence policy outputs and measure them (Ingold and Christopoulos 2015: 17). In accordance with Ingold and Christopoulos, the existence of different drivers and pathways to the adoption of the tobacco advertisement bans is assumed. No single factor is responsible for the policy output; rather, it is different constellations of influential aspects that need to be considered. What is crucial to recognize furthermore is the detection of those factors not only in the area of political decision-making but also before that, namely at the stage of agenda-setting: In four cantons the adoption of advertisement bans was never even discussed. It is likely that different dynamics were at play in this phase of policy-making than later on, which is why it will be analysed separately from the stage of decision-making. Herweg et al. (2015) provide a useful tool for such a split analysis: By extending and adapting Kingdon s Multiple Streams (1995) they develop an explanatory framework for the entire policy-making process. This promising approach until today has only been tested sporadically (Zohlnhöfer et al. 2015; Zohlnhöfer 2015) and never in a comparative way (Zohlnhöfer 2015: 22). The present study therefore poses a novelty from a theoretical point of view as well. I use the fact that the Swiss cantons are fully fledged yet comparable political systems due to their common national context to single out relevant causes and causal combinations for the discussion and adoption of tobacco advertisement bans (Vatter 2002). The method of Comparative Qualitative Analysis (QCA) is considered to be ideally suited for this endeavour: It allows the comparison of a multitude of cases and focuses not only on individual explanations, but on necessary and sufficient configurations (Berg-Schlosser et al. 2009). A QCA shall thus be conducted for the agenda-setting and the decision-making process each. The period of investigation starts with the Federal Court s sentence in 2002 and ends in 2009, when the last advertisement ban was adopted in Ticino. In the next part I will introduce the phenomenon of interest: tobacco advertisement bans. After, Kingdon s Multiple Streams Framework (MSF) and the modification carried out by Herweg et al. (2015) will be presented. Following up, the explanatory model is to be applied to the case and the hypotheses are derived. After a brief presentation of the method of QCA I will report 3

4 the results of the analysis. The paper ends with a conclusion that discusses implications of the findings. Tobacco advertisement bans From a political science perspective the advertisement bans analysed in this paper constitute a public means to effect social change (Vedung 2000: 122). The ultimate goal of the bans is to reduce the prevalence of smoking in a state s population; in order to achieve it, the tobacco industry is targeted with a law that impedes their possibilities of recruiting new consumers. In contrast to other instruments of public policy like the distribution of information or financial incentives, a ban establishes maximal obligation: Infringements will be prosecuted. Nevertheless, the scale of a ban may be varied in terms of its reach. A ban is more severe if it forbids the sponsoring of cultural events than if it only mentions billboard advertisements. This is exactly what can be observed in the case of the Swiss tobacco advertisement bans: While in some cantons a total ban on every kind of tobacco advertisement is imposed, others restrain their bans to specific areas like billboards on public buildings and cinema spots. It therefore makes sense to consider the cantonal advertisement bans as specifications of a general policy instrument, which can be configurated individually according to the reach of the ban that indirectly translates into the stringency with which its addressee, the tobacco industry, is to be treated. 4 To achieve an elaborate rating of the different bans, an index was created. A policy proposal gets assigned one point for every area of application it mentions 5 ; the higher the number on the index, the bigger the reach of the ban. Two areas of application are weighted doubly, as they pose in fact a more restrictive and general ban than the others. They concern advertisements in the public area and advertisements in places that are visible from the public area. A proposal targeted at prohibiting tobacco advertisement in general is assigned the maximum value. The created index ranges from 0 to 9. Bans that achieve a value of 5 or below are considered to be rather weak, whereas bans exceeding that threshold constitute rather strong bans. This rests on the observation of previous studies that an advertisement ban comprising only parts of the potential advertisement channels is hardly a ban at all: it is very easy for tobacco producers and sellers to redirect their 4 A similar gradation of a policy instrument s stringency has also been performed by Knill et al. (2012), who compared clean air regulations across different nation states. 5 Those areas are namely: public buildings, sports events, cultural events, events for young people under the age of 18, cinemas. 4

5 promotion (Werner et al. 2016: 20). The qualitative distinction between weak and strong bans will become important for the QCA below. A multiple streams approach to agenda setting and decision making John Kingdon established in his seminal book Agendas, Alternatives and Public Policies a framework of agenda-setting in public policy-making under conditions of ambiguity (1995). Ambiguity denotes a state of having many ways of thinking about the same circumstances or phenomena (Feldman 1989: 5, cited after Zahariadis 2014: 26). This is a concept which can be applied to virtually any topic in public policy nowadays: The banning of tobacco advertisement for example can be treated as a health issue when thinking about the risks that smoking poses for an individual s health. It can also be framed in a way that emphasises questions of commercial freedom however: May the state prohibit a tobacco manufacturer from promoting his products, or should it let the market have its way? In order to understand at which point in time a certain topic pops up on the public agenda and what form it takes when it does so, Kingdon assumed the existence of three independent streams that float through the political space (1995: 87). The so-called problem stream encompasses all societal and environmental constellations that could potentially be interpreted as problematic and worthy of being politically tackled. The policy stream in turn consists of proposals by scientists and other experts about how the state could get active in different policy areas (Kingdon 1995: 117ff.). The third political stream depicts the power relations between actors of the political process: Which parties are dominant in the executive and legislative branch, are there any interest groups that have the potential of influencing public policy etc. For a certain topic to appear on the political agenda, the three streams have to converge in a specific way: An individual entrepreneur needs to take the initiative and actively couple them with the goal of pushing a policy proposal he or she personally favours on the agenda (Kingdon 1995: 179). Translated into non-metaphorical language this would mean that an individual successfully manages to present a specific proposal as the optimal solution for an existing problem, and that the current political elite is receptive enough to discuss this suggestion in the formal decision-making process. The time frame during which such a coupling of the streams can be achieved is limited (Kingdon 1995: 165f.). This results from the contrast between the infinite list of issues that might be worthy of political attention and the limited capacity of the political system and its actors to process information (Kingdon 1995: 184f.). Kingdon postulates that some kind of event has to trigger the opening 5

6 of such a time frame, also known as window of opportunity (Kingdon 1995: 165f.). This event raises the general attention for an issue and thereby offers the entrepreneur a platform for presenting his new approach to the relevant actors. After a certain time the effect of the event starts to wear off; the window of opportunity closes and other topics are prioritized. Kingdon s framework, which became famous under the name of Multiple Streams Approach (MSA), has met with a lot of success in the scientific community (Zahariadis 2014: 25). It also has been frequently expanded to not only explain agenda-setting, but also the process of decision-making (Zahariadis 1992: 359). As Herweg et al. (2015: 444) correctly observe though, an unmodified transfer of the theoretical model to this stage of policy-making or even a fusion of the agenda-setting and the decision-making process into one unit of analysis bears the risk of information loss and misinterpretation. If one observes for example the lack of a policy change in a system it is unclear whether this is due to a failure in bringing a policy proposal to the political agenda or to missing majorities in the decision-making stage. Instead of viewing the two stages as one overarching coupling of the streams Herweg et al. therefore suggest a duplification of the coupling process, with minor adaptations for the second stage of decision-making (Herweg et al. 2015: 444ff.). They posit that once a policy proposal has reached agenda-status i.e. when the first process of coupling has been successfully completed the window for the second coupling opens automatically: Politicians, interest groups and other relevant actors do not want to miss the opportunity of influencing a policy that is up for decision and start negotiating immediately (Herweg et al. 2015: 444f.). The political stream and the entrepreneur become the crucial factors now, as it is the majority in parliament which will decide on the policy s fate (Herweg et al. 2015: 445). With the current distribution of power between the parties in mind the entrepreneur tries to forge a coalition for his or her pet proposal (Herweg et al. 2015: 445f.). If his efforts are successful, the policy is formally adopted. In what follows, I will adopt Herweg et al. s modification of the multiple streams framework. As the analysis at hand focuses not only on one case but on the totality of the Swiss cantons, several adaptions will further have to be taken into account. They will be described in detail in the next section, where the theoretical framework is transferred to the empirical cases. 6

7 Deduction of hypotheses Based on the theoretical elaborations above I will now apply the explanatory model to the tobacco advertisement bans in Switzerland. The two analytical steps are treated separately. A compilation of the cantons scores on the various elements is included in appendix A. Agenda setting stage Before a political discussion can be held on the advantages and disadvantages of adopting a tobacco advertisement ban, the topic needs to be put on a canton s agenda. The first coupling of the streams will therefore be defined as successfully completed as soon as a ban appears in parliament (BANAG 6 ). It may take the form of a drafted bill by the government or of a parliamentary proceeding, namely a motion or a postulate. The window of opportunity, which allowed the tobacco advertisement bans to appear on the cantonal agendas, was opened by a sentence of the Swiss Federal Court. On 28 May 2002 its judges decided that a correspondent law issued by the cantonal parliament of Geneva was compatible with the national constitution and could therefore be implemented 7. A close look on the subsequent parliamentary debates in the other cantons shows that a lot of policy promoters explicitly referenced this sentence as a supporting argument 8. The decision also received a great deal of attention in the national media (Felber 2002; Inderbitzin 2002) as it had been unclear up to that point whether this topic should be dealt with by the national entity or by the federal units. The Federal Court made it possible for other cantons to follow Geneva s example. Why of all cantons it was Geneva to first deal with such a law becomes clear when one considers its unique embedment in international politics: Since the foundation of the United Nations Organization (UNO) and its placement in Geneva shortly after the Second World War many other international actors settled there (Heimberg 2011). Among them is the World Health Organization (WHO), a long-standing advocate of advertisement bans 9. Mavrot (2015) notices striking parallels between the Geneva policy proposal and guidelines of the WHO. A report published by the WHO one year before is also mentioned during the parliamentary discussion by one of the 6 The abbreviations used for the analysis in a later stage are reported here in italic letters. 7 BGE 128 I Cf. e.g. Protocol of the Cantonal Parliament of Berne 2006: 419; Protocol of the Cantonal Parliament of Zurich 2003: Cf. e.g. World Health Organization (1999: 73). 7

8 initiators of the new law (Bulletin du Grand Conseil Genève 2000: n.p.). It can thus be assumed that the local policy community was in contact with the WHO or at least informed about its endeavours in this area of health prevention. The political stream will be composed of elements that depict the prevalent political interests concerning tobacco advertisement bans in a canton. On the one hand, the partisan composition of parliament and government matter here (LEFTP and LEFTG). Various studies have demonstrated that leftist parties 10 in Switzerland often engage in the promotion of public health issues (Sager and Rielle 2013; Trein 2013: 6). Their presence in a canton s political institutions is therefore of vital importance for the introduction of the bans to the official agenda. On the other hand, organized interests outside of the institutions have to be taken into consideration. First of all, we need to think of the actors directly addressed or negatively affected by the advertisement bans: The tobacco producers, cigarette and cigar manufacturers as well as companies mainly engaging in tobacco trade (TOB). Even though tobacco production is on the decline in Switzerland (Schmutz 2014), some 300 farmers still rely on this crop 11. What is more, the biggest cigarette and cigar producers are highly present in Switzerland: The three global concerns Philip Morris International (PMI), Japan Tobacco International (JTI) and British American Tobacco (BAT) all maintain a factory and/or important administrative centres here, and the two famous companies Villiger Tabak and Burger Söhne AG (producer of Dannemann cigars) are even of Swiss origin 12. An extensive document analysis has shown that in the past the tobacco industry has indeed tried to exercise its influence upon Swiss politics (Lee and Glantz 2001). For the present study, similar documents were unfortunately not available. The Swiss Federal Statistical Office (SFSO) provides however data on full-time equivalent employees in tobacco-related branches for each canton 13. This measure serves as a proxy for the general weight of tobacco interests: Not only does it depict the number of people who will potentially oppose the law, but it also shows the economic importance of tobacco production and trade in 10 For the study at hand, the following parties were considered to be leftist : SPS, GPS, CSP, FGA, POCH, PSA, Sol. The Christian-Social Party (CSP) usually is not considered to be leftist; as they list on their website the targeted prevention of tobacco, alcohol and drug abuse as one of their goals, they were included in this measurement. Cf. accessed on accessed on Cf all accessed on accessed on

9 a canton. At the agenda setting stage one can imagine that the lobby itself will not be active yet, but that politicians dealing with the subject will already acknowledge it as potential opponent and consider the danger of harming a source of employment. The lobby potentially fighting for the introduction of an advertisement ban in a canton can be measured in a similar way: The SFSO also provides data on the full-time equivalent employees in organisations promoting health, like the league against lung cancer for example 14. At this point it is important to take into account a special feature of the Swiss political system: In a lot of cantons close ties exist between the public administration and those health interest groups due to extensive collaboration especially in policy implementation, but also in policy formulation. Especially when there is at least one person in the administration assigned with the task of improving tobacco prevention, they therefore have a special channel to the policy process. It is furthermore possible to assume that such a person or office would put additional effort into the promotion of advertisement bans as this constitutes a fundamental part of their raison d être. The measure of health organizations in a canton will thus be complemented by a dummy indicating the presence or absence of such an administrative entity. This combined condition will serve as an indication for the strength of a health lobby in a canton (HEALTH). The data for this stem from a cantonal survey conducted at the beginning of the period of investigation (Sager et al. 2003) and from Sager and Balthasar (2013). A theoretical element which will not be differentiated for each single canton is the problem stream. The fact that tobacco consumption is harmful to one s health has dominated the international discourse for several decades. The Swiss state too passed the first regulations that restricted the tobacco industry as early as 1964 (Trein 2013: 8). The problem can thus be considered to have been pressing independently of regional preconditions. Information gathered by the Swiss health survey corroborates this decision: Although there is some variation visible in the cantonal prevalence of tobacco addicts, the gap between the canton with the lowest and the canton with the highest rate amounts to merely 10 percent. The policy stream, i.e. the solution proposed to the problem of tobacco consumption and its damaging effects on health, is to be held constant in the first part of the analysis as well. As has been shown at the beginning of this paragraph, the politicians from Geneva drew heavily on the accessed on

10 proposal designed by the WHO. Mavrot (2015) by means of an in-depth process analysis convincingly demonstrates that the advertisement bans were first transferred from the international level to Geneva and then spread among the cantons. It will thus be assumed in the first step of analysis that in principle the policy proposal of banning advertisement on tobacco products was known to all the relevant actors in Switzerland. This claim is further supported by the fact that a few years before the Swiss people had already been confronted once with the proposal of a comprehensive advertisement ban. In 1993, a popular initiative on the topic was rejected at the polls (Dubach 2010). Lastly, the entrepreneur as a catalysing element will be omitted from the analysis of the agenda stage too. This has to do with the special nature of the policy examined here: On the one hand, Cairney et al. correctly observe that the introduction of advertisement bans is but one small step in a big policy change on the international level from tobacco support towards tobacco prevention. This profound alteration in perception and political behaviour has taken place over decades since the 1950s and has always proceeded incrementally: A multitude of introduced instruments, such as educational provisions, the interdiction of smoking in restaurants or the ban of commercials in TV and radio all serve the same primary goal of reducing overall tobacco consumption (Cairney et al. 2012: 221ff.). As Mintrom and Norman (2009: 650) highlight, entrepreneurs become especially important when a radical innovation or turn in current behaviour is to be conducted. Incremental adaptations can mostly be handled by the system without this special impetus. On the other hand, one can argue that there wasn t even a need for entrepreneurs when the advertisement bans were to be set on the cantonal agendas. Mintrom (1997: 740f.) noticed in his study on the role of entrepreneurs in the diffusion of policy innovations that sometimes topics get on the agenda without an entrepreneurial effort. Legislators or members of the government at that might introduce a new policy as part of their daily routine, which is nicely illustrated by this citation of a politician that was interviewed by Mintrom: I think you have some people that are just copying. They ve heard about this idea and they say, Let s have one of these too. (Mintrom 1997: 741). For federal states in general and Switzerland especially this is a common practice which has been empirically illustrated several times already (cf. for example Schaltegger 2004; Gilardi and Füglister 2008). Accordingly, the entrepreneur is not considered to be a decisive part of the explanation of tobacco advertisement bans in the cantons. 10

11 Hypothesis for the agenda-setting stage Following the considerations of the previous sections I will test whether the presence of the political stream in a canton can be counted as a sufficient condition for the appearance of a tobacco advertisement ban on the canton s political agenda. The political stream is divided into the three sub-conditions of strong leftist parties in parliament, strong leftist parties in government and the combination of the presence of health organizations or of an administrative unit responsible for tobacco prevention. The hypothesis can also be expressed in a formalized manner 15 : H1: LEFTP*LEFTG*tob*HEALTH BANAG Decision-making stage In accordance with Herweg et al. (2015), the window for the second coupling process is regarded to open automatically as soon as a policy proposal has reached the official agenda (cf. above). Due to this reasoning, it will not be included in the analysis. The same holds for the problem stream, which again is interpreted to be present in all the cantons because of the global consciousness that smoking is a serious health hazard. The policy stream at this stage depicts the solution under discussion, that is to say the end product of the first coupling process. As has been illustrated above, the advertisement bans vary in their reach of application (REACH). During decision-making, a severe formulation of the policy proposal might pose an obstacle to its adoption; the larger the reach of application, the more there is to lose for the policy s opponents. An encompassing advertisement ban also constitutes a bigger deviation from the status quo, which from an institutionalist perspective always obstructs policy change (Tsebelis 2002). In a similar vein, a finding from previous research on tobacco advertisement bans in Switzerland will be included here: While examining the parliamentary debates on tobacco advertisement bans in the French-speaking cantons, Mavrot et al. (2016) noticed that a decisive factor concerning the adoption or rejection of the law was its (non)-connection to an analogous ban on alcohol advertisement. Such a link to alcohol prevention policy turned out to be a hindering factor for the policy proposal since the zone of attack 15 The applied notation corresponds to the standards most often used in QCA analysis. Uppercase letters indicate the presence of a phenomenon, lowercase letters indicate its absence. The asterix (*) denotes a simultaneous occurrence of two phenomena (intersection of conditions), the plus-sign would mean that at least one of the two phenomena is present (union of conditions). An arrow that points to the right ( ) represents sufficiency of the solution on the left-hand side for the concept on the right-hand side, an arrow to the left ( ) would denote necessity. 11

12 for policy adversaries was enlarged. The production of alcoholic beverages is an important cultural value of the region as lots of cantons have their own wine-production. It was therefore easier for opponents to mobilize resistance against the new law (Mavrot et al. 2016: 16). A similar mechanism could be relevant for the rest of Switzerland as well, since several other regions such as Biel or Zurich maintain vineyards. The connection of the tobacco advertisement bans with alcohol advertisement bans (ALC) will therefore be considered to be a second important element of the policy stream in the decision-making phase. The political stream, which gets ascribed a predominant role at this stage of policy-making by Herweg et al. (2015: 445), is similarly constructed as in the first coupling process. Again, the strength of leftist parties (LEFTP), the presence or absence of the tobacco lobby (TOB) and of health organizations respectively the existence of an administrative unit with the task of promoting tobacco prevention (HEALTH) are part of it. For this stage, the partisan composition of government will not be included anymore however: As soon as a legislative draft has entered the parliamentarian arena the government loses its influence over it. It has passed from the executive to the legislative power, with the former becoming mere bystanders and advisers in the process that follows. It is the proportion of supporters in parliament that now becomes a decisive factor for the success or failure of a law. Similar to the first coupling process, the entrepreneur will not be treated as a separate feature. The reasoning for this is now different though: Herweg et al. (2015: 445f.) explain that during the decision-making stage the entrepreneur will most likely be an individual inside the formal governmental system with an elected leadership position. Even if he or she resides on the outside of it, an internal entrepreneur will have to help gathering the majorities needed for the adoption of the supported proposal. Being formally elected, the entrepreneur is a member of one of the parties and as such part of the partisan balance of power, which is depicted in the political stream. As it is very difficult for a partisan entrepreneur to strike a deal when the balance of power clearly outweighs him or her, the partisan composition of parliament already included in the political stream is considered to be enough of an indicator for a proposal s chances to pass the vote. The result emanating from the second coupling process finally will be operationalized analogously to the one from the first step of analysis. The final legislation drafts that were adopted by the cantonal parliaments serve as the basis for this measure (BANAD). 12

13 Hypothesis for the decision-making stage Subsequent to these reflections, the hypothesis for the second step of the analysis may be formulated. Regarding the policy stream, a proposal with wide reach that is linked to an analogous ban on alcohol advertisement constitutes a barrier to its passing the vote. If a ban is to be adopted, the policy stream should therefore be weak rather than strong. As in the first analytical step, the strength of left parties in parliament is expected to be supporting the adoption of an encompassing advertisement ban while the presence of the tobacco lobby constitutes a hindering factor. The third element of the political stream, which measures the presence of health organizations or of an administrative unit responsible for tobacco prevention should also support the advertisement bans. Again, the hypothesis can be expressed in a formalized manner: H2 reach*alc*leftp*tob*health BANAD The method to be applied: Qualitative comparative analysis QCA was developed by Charles Ragin (1987) to address the famous methodological problem of many variables, few cases. The application of set theory and Boolean algebra allows the reduction of complex constellations into concentrated solutions. However, a fundamental knowledge about the single cases is needed in order to achieve valid results (Rihoux and Meur 2009). Characteristics of the cases ( conditions ), that from a theoretical point of view are thought to be relevant for the explanation of a certain feature ( outcome ), are conceived of as sets. To use an example from the study at hand: A canton may be defined as member of the set cantons with strong leftist parties in government if the absolute majority of elected executive officials are from parties such as the SPS or the GPS. The canton will be counted as non-member of the set conversely, if none of the officials stem from leftist parties. The special variant of QCA that is to be used in this analysis, called fuzzy-set QCA, allows a further distinction for cases that seem more ambivalent: If e.g. in a cantonal government three of seven members are leftist, then the canton will be counted as rather in the set of cantons with strong leftist parties in government, but not fully in. The process of assigning set membership scores to the cases is called calibration (Ragin 2009: 89ff.). A documentation of the calibration for this analysis may be found in appendix B; the cases final fuzzy scores are displayed in appendix C. 13

14 After calibration, the cases set membership scores are compared with regard to the designated outcome in the study at hand the introduction of tobacco advertisement bans on the agenda respectively the adoption of the bans in parliament. The goal is to find systematic patterns across the cases: Does a certain condition or a combination of conditions always appear when the outcome is present as well? Then it is deemed to be a sufficient condition. Or does the outcome only show up when a condition or again a combination of conditions is fulfilled? Then this (set of) condition(s) might be necessary for the outcome. By comparing the cases in such a way, possibly causal connections between conditions and the outcome may be discovered (Rihoux and Meur 2009: 44ff.; Ragin 2009: 99ff.). To substantiate those findings, the researcher then has to go back into the individual cases and look for the exact mechanisms that might be responsible for this set-theoretic result (Schneider and Wagemann 2010: 400f.). The explanations above obviously constitute a very rudimentary introduction to QCA. It is possible here to refer to two qualified introductory texts if further information is asked for: Rihoux and Ragin (2009) and Schneider and Wagemann (2012). What should have been made clear by this brief description though is the fundamental advantage QCA has when it comes to applying Kingdon s Multiple Streams to empirical material: It allows the detection not only of single causal connections, but also of causal complexity. In essence, Kingdon s model is a configurative one: For a policy proposal to appear on the political agenda or to be adopted by parliament, as it is the case in Herweg et al. s modification a combination of the streams with the policy window and the entrepreneur is needed. The presence of a problem alone for example does not guarantee its treatment by the political system. QCA helps to figure out whether in empirical reality too it is a combination of factors that are responsible for the results of policymaking or not. Results Agenda setting stage During the analysis of smoking bans appearing on the cantonal agendas, no necessary conditions could be detected. The sufficient solutions for the positive outcome ( strong ban appearing on the cantonal agenda ) and the negative outcome ( no ban or weak ban appearing on the cantonal agenda ) are depicted in table 1 and table 2 respectively. 14

15 Table 1: Analysis of sufficiency for the outcome BANAG 16 Solution: LEFTP*leftg*HEALTH BANAG Single case coverage: SO, BL; AG Solution consistency: Solution coverage: Italic: deviant case consistency in kind (membership > 0.5 in path but not in outcome). The raw consistency threshold was set at Cases separated by semicolons belong to different truth table rows. Directional expectations: LEFTP BANAG, LEFTG BANAG, tob BANAG, HEALTH BANAG. Figure 1: Plot of the sufficient solution for BANAG 16 Abbreviations used: BANAG = strong tobacco advertisement ban appears on the political agenda; LEFTP = strong leftist parties in parliament; LEFTG = strong leftist parties in government; TOB = strong tobacco interests present; HEALTH = strong health lobby (health organizations or administrative unit promoting tobacco prevention) present. 15

16 For the appearance of a strong tobacco advertisement ban on a canton s agenda the simultaneous occurrence of strong leftist parties in parliament, weak leftist parties in government and a strong lobby for the promotion of tobacco prevention is deemed to be a sufficient combination of conditions. The QCA parameters of fit forbid a substantive interpretation of these results though: While the solution consistency reaches an acceptable value of 0.805, the plot of the solution (cf. figure 1) reveals a deviant case in kind that contradicts the statement of sufficiency. The canton of Basel-Land exhibits the combination of conditions in question and still has a rather weak advertisement ban on the agenda consisting only of the ban of billboard advertisement in the public area and in places that are visible from the public area as well as on public buildings. The sponsoring of cultural or sports events, of events for young people under the age of 18 or the running of an advertisement spot in cinema are not included. While the deviation of one canton could potentially be explained by an in-depth case analysis, the low over-all coverage of the sufficient solution (0.363, only three cantons covered) is another piece of evidence for the lack of explanatory power of the applied model. Going back to the theory and the cases seems appropriate in order to get a better understanding of the phenomenon. Table 2: Analysis of sufficiency for the outcome banag Solution: LEFTG*TOB + LEFTP*health + leftp*leftg* HEALTH banag Single case coverage: ZG; NE; FR, BS, VD, JU TG; SH; NE OW; ZG Consistency: Raw coverage: Unique coverage: Solution consistency: Solution coverage: The raw consistency threshold was set at Cases separated by semicolons belong to different truth table rows. Directional expectations: leftp banag, leftg banag, TOB banag, health banag. 16

17 Figure 2: Plot of the sufficient solution for banag The sufficiency analysis of the negative outcome, i.e. the appearance of only a weak tobacco advertisement ban or no ban at all on the cantonal agenda, yields more interesting results in terms of model quality. The solution consistency is higher (0.913) and no case contradicts the statement of sufficiency (cf. figure 2). The overall coverage too this time indicates a relatively good explanation of the phenomenon examined (0.715). Three paths can lead to the lack of a strong ban on the cantonal agenda (cf. table 2). What immediately strikes the eye is the unexpected role leftist parties in government assume in two of the three paths. An executive with a strong left seems to hinder rather than further the introduction of tobacco advertisement bans with a wide reach to the political arena. An explanation for this finding might be the responsibility a politician has to assume as soon as he or she is elected into government: Especially in Switzerland, where the executive is usually a coalition of three or more parties, it is often not possible anymore to maintain the ideological position that previously dominated a politician s behaviour. Decisions are taken on a consensual basis, which leads to a more pragmatic interpretation of a jurisdiction s current problems and needs. Particularly the first path, where the strong left in government is combined with a strong representation of tobacco interests, it is conceivable that economic reflections prevented the adoption 17

18 of a clear stance against tobacco advertisement. The topic seems to have been avoided or kept at a low scale so as not to endanger any jobs. Why the support of the left in government by a strong health lobby as depicted in path three is not sufficient for a strong advertisement ban on the agenda remains unclear. A further investigation into the two cases subsumed under this path could lead to a better understanding. The second solution path denotes the presence of a strong left in parliament together with the absence of a strong health lobby as a sufficient condition for weak or no bans on the cantonal agenda. Here it looks as if even though the general political context was favourable towards the advertisement bans, the missing respectively weak promoters of the policy turned the balance against a proposal with a wide reach on the agenda. With regard to the hypothesis formulated above for the agenda-setting stage, especially the configurational expectations have to be rejected: The combined presence of the political stream s components was not detected to be a sufficient condition for the appearance of strong tobacco advertisement bans on the agenda. Focusing on the single elements the results are also mixed. While the presence of tobacco interests in the model behaves in accordance with the underlying theory it does not appear in the sufficient solution for the positive outcome and is part of a sufficient path for the negative outcome, a strong left wing in parliament only partly does so. Depending on the specific configuration with other elements, it may also be an obstacle for the discussion of wide-reaching advertisement bans in a canton s political institutions. The same conclusion needs to be drawn concerning the presence of a strong health lobby. A strong left in government finally apparently has acted completely against the voiced expectations. As has been mentioned above, it might however be necessary to further refine the theoretical model based on substantial knowledge of the cases, since the parameters of fit for the analysis of the positive outcome turned out rather low. Decision-making stage While analysing the decision-making stage on tobacco advertisement bans, two necessary conditions for the positive outcome a strong ban adopted by the cantonal parliament were found: First it is necessary that the proposal entering the parliamentary discussion already displays a wide reach, and that there are no tobacco interests present in the canton 17. We can learn from 17 Parameters of fit for REACH*tob: Consistency = 0.926, Coverage = 0.808, Relevance of Necessity =

19 this that apparently no proposal entering parliament ever left it being bigger: If modifications were made, they came in the form of reduction or abandonment. For a canton to adopt a strong ban, the ban had to be strong from the start. Even if this was the case however, the chances of ban adoption vanished when a significant amount of tobacco producers, manufacturers and traders was based in the canton. It seems that the lobby has a big influence on the decisionmaking stage. Whether this influence is based on active lobbying or on mere presence remains unclear; even though it is imaginable that a tobacco producer would get politically active as soon as he or she hears about the law proposal in the media, there is also the possibility that politicians weigh in this economic factor more when a definite decision is about to be made than when a subject is only up for discussion. As lobbying is often performed in a covert way, it is likely that we will not obtain more information on the matter. For the negation of the outcome no single condition fulfilled the statement of necessity. The sufficient solutions for the positive outcome ( strong ban adopted by the cantonal parliament ) and the negative outcome ( no ban or weak ban adopted by the cantonal parliament ) are depicted in table 3 and table 4 respectively. Table 3: Analysis of sufficiency for the outcome BANAD 18 Solution: REACH*tob * ( alc*health + ALC* HEALTH + LEFTP* HEALTH ) BANAD Single case coverage: SG UR, AR; BE SO; BE Consistency: Raw coverage: Unique coverage: Solution consistency: Solution coverage: The raw consistency threshold was set at Cases separated by semicolons belong to different truth table rows. Directional expectations: LEFTP BANAD, tob BANAD, HEALTH BANAD, reach BANAD, alc BANAD. 18 Abbreviations used: BANAD = strong tobacco advertisement ban adopted by parliament; LEFTP = strong leftist parties in parliament; TOB = strong tobacco interests present; HEALTH = strong health lobby (health organizations or administrative unit promoting tobacco prevention) present; REACH = strong tobacco advertisement ban entering the parliamentarian discussion; ALC = connection of the discussed advertisement ban to an alcohol advertisement ban. 19

20 The minimization of the truth table for the positive outcome reveals three possible solution paths (cf. table 3). Figure 3 demonstrates that there is no case contradicting the statement of sufficiency; the overall consistency is at 0.876, the solution coverage amounts to Figure 3: Plot of the sufficient solution for BANAD The relevance of a strong law proposal and the absence of the tobacco lobby is further emphasized by the sufficient solution (table 3): They are part of all three solution paths and can therefore be factorised. The first path is characterized by a complete absence of actors that might push the proposal: Neither the political landscape being composed of parties in parliament and interest group organizations nor the public administration are shaped favourably. The absence of the contra-lobby seems to compensate for this however. Path two exhibits a particularly strong policy proposal, with a big reach of the tobacco advertisement ban as well as the combination with an alcohol ban. The combination with a strong health lobby seems to be an important factor here: Actors are needed that can substantiate the proposal with arguments based on evidence in order to get it passed. 20

21 The last path features all of the political stream s elements in the form they were expected to further the cause of tobacco prevention. If the political and administrative landscape are receptive for the advertisement bans, this is a sufficient condition for them to pass the vote. Table 4: Analysis of sufficiency for the outcome banad Solution ALC*TOB + reach*alc + reach*health banad Single case coverage: ZG; NE; BS, VD; LU; AG NW, GR; ZG; TG; ZH, BL; NE; BS, VD SH; TG; NE Consistency: Raw coverage: Unique coverage: Solution consistency: Solution coverage: The raw consistency threshold was set at Cases separated by semicolons belong to different truth table rows. Directional expectations: leftp banad, TOB banad, health banad, REACH banad, ALC banad. The solution for the negative outcome also consists of three paths with an overall consistency of and coverage of (cf. table 4). The first path features the presence of a strong tobacco lobby and a tobacco advertisement ban that is complemented by an identical ban on the promotion of alcoholic beverages. It seems that this enlarged proposal offers a good opportunity for its opponents to advocate its rejection, or that they are even joined in their efforts by a second lobby from the sector of alcohol production. Either way, the combination proves fatal for the advertisement bans. The second and third path both contain a weak ban proposal entering the discussion, which for the politicians might constitute too little a measure to have any significant effect. If an advertisement ban only covers a small part of the possible channels, then it might as well be abolished altogether. In the second path this weak proposal merges with an alcohol advertisement ban, whereby the pool of possible opponents increases. The third path in contrast displays a combination with weak or no mentionable health proponents; there simply is nobody to lobby for a law that only is a weak statement anyway. 21