Introduction
Since its foundation, the European Union (EU) has deepened the integration of its member states through legally binding regulations, which needed to be implemented by member states, through nonobligatory guidelines, and through softer modes of governance. One binding regulation is the EU directive on tobacco advertising. Given that some of its member states had banned it for quite some time, in 1998, the EU passed a directive aimed at phasing out all tobacco advertisements and sponsorship by 2006. Following a challenge by Germany and other countries, the European Court of Justice (ECJ) declared the directive invalid. In 2003, a less restrictive directive (2003/33/EC) followed, against which some member states, including Germany, sought annulment in the ECJ, but this time the court refused to void the directive (Duina and Kurzer Reference Duina and Kurzer2004; Studlar Reference Studlar2006a, 387).
Although the tobacco advertising directive was legally binding for EU members, an interesting and important question is to what extent this regulation, and possibly other EU directives, indirectly affected nonmember states of the EU, such as EU neighbours and those who hold many bilateral treaties with EU nations. It is plausible to assume that there is some influence (Schimmelfennig Reference Schimmelfennig2009), as the EU’s unique governance attracts countries in its periphery. In the best case, this could lead to more innovative policies, and in the worst case it could lead to conflicts within these countries. Switzerland has a special relation with the EU because of bilateral treaties that grant it full access to the European Economic Era (EEA) but do not entail automatic implementation of all regulations that bind full EEA members. Furthermore, Switzerland has relatively many “member states” (26 cantons), which differ culturally and politically and have considerable autonomy in many areas; this makes the country an even more interesting case of indirect influence of the EU, as we can observe subnational governments.
Between 1998 and 2010, 15 cantons restricted tobacco advertising; others have not done so until now. So far, we know neither what explains differences in cantonal adoption of advertisement bans nor whether there has been any research on the questions regarding whether these regulations are related to the European dimension. If we were to find a relationship between European politics and the restriction of tobacco advertisement by Swiss cantons, such a finding would be highly interesting for policymakers and researchers alike, as it would demonstrate that policy ideas related to EU directives might – even indirectly – bypass national politics and reappear on the political agenda of subnational governments.
This article analyses why there are differences between cantons regarding the adoption of tobacco advertising bans. The main argument of the article connects popular support for European integration to cantonal adoption of tobacco advertisement restrictions. Theoretically, the article starts from the “social contagion model” in the policy diffusion literature (Pacheco Reference Pacheco2012), as well as research on economic voting (Duch and Stevenson Reference Duch and Stevenson2008) and electoral politics (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Jacoby Reference Jacoby1994; Stevenson Reference Stevenson2001). The article uses the mentioned literature to develop an argument that connects the European dimension of tobacco advertising restrictions to their adoption by the governments of the Swiss cantons. Precisely, the article argues that voters’ support for more European integration serves as a signal for policymakers. The more the cantonal electorate supports deeper European integration, the more policymakers are inclined to ban tobacco advertising restrictions. This assumption is plausible for the case of tobacco advertising bans in Switzerland as there is no clear information about the effectiveness of this instrument. Therefore, policymakers need to use signals, that is, cognitive shortcuts, to estimate the political risk of such a reform proposal. Voters’ support for European integration is a powerful signal for tobacco advertising restrictions given the European dimension of the topic, as it is a salient issue that politicises the Swiss population and as Swiss media have framed tobacco advertisement bans as a topic related to Europe.
To empirically support this claim, the article uses statistical analysis, especially event-history analysis (EHA) based on logit models. The results support the argument and are robust when control variables and measurements for time dependence are inserted into the analysis.
This article contributes to the literature on tobacco control by examining anti-smoking policies in Switzerland, notably tobacco advertisement bans. This is important research, because there has been little inquiry into tobacco control in Switzerland.Footnote 1 What is more, although the empirical scope of the article is restricted to tobacco advertisement restrictions in the Swiss cantons, the results also contribute to the political science literature more generally. First, the findings add to the policy diffusion literature (Walker Reference Walker1969; Simmons and Elkins Reference Simmons and Elkins2004; Braun and Gilardi Reference Braun and Gilardi2006; Dobbin et al. Reference Dobbin, Simmons and Garrett2007; Shipan and Reference Shipan and VoldenVolden Reference Shipan and Volden2008; Gilardi Reference Gilardi2013; Maggetti and Gilardi Reference Maggetti and Gilardi2016), notably to the “social contagion model”, according to which policymakers decide based on voters’ support of policies abroad whether to put them into place at home (Pacheco Reference Pacheco2012). This article underlines that – next to the electorate’s approval of the policy as such – voters’ support for the country or jurisdiction that serves as a model for a certain policy influences policymakers’ decision to adopt it. The country of origin of a given policy influences the framing of the policy solution in the democratic discourse. Thus, policymakers tend to avoid pursuing reforms that originate in countries with a bad reputation as this decreases the acceptance of the reform for political reasons. Further research could apply this argument to other examples of policy diffusion.
Second, the article contributes to our understanding of Swiss politics and the impact of the EU on them. A large body of literature compares democratic structures and policy outputs at the subnational level in Switzerland (Vatter Reference Vatter2002; Vatter and Stadelmann-Steffen Reference Vatter and Stadelmann-Steffen2013). The findings of this article could be interesting to researchers who work on subnational policymaking, such as integration policy (Manatschal Reference Manatschal2011). Popular support of more European integration might affect the adoption of “Europeanised” policies at the cantonal level lead to different patterns in cantonal policy outputs. Second, this research teaches us that differences in support for more European integration between regions might affect countries that are close to the EU but are not actual members, because the political and social model of the EU (Schimmelfennig Reference Schimmelfennig2009) affects other countries that are nearby, including at the subnational level.
Diffusion, tobacco control policies and mass attitudes
The literature on tobacco control policy has emphasised a number of factors that are important to differences in the adoption of tobacco control policies. Notably, recent contributions have pointed to differences in patterns of adoption of tobacco control policies among nation states in Europe (Cairney et al. Reference Cairney, Studlar and Mamudu2012). What is more, researchers identified corporatist institutions (Studlar et al. Reference Studlar, Christensen and Sistari2011, 738) and a strong tobacco industry (Bornhäuser et al. Reference Bornhäuser, McCarthy and Glantz2006; Studlar Reference Studlar2006b; Grüner et al. Reference Grüner, Strünck and Gilmore2008; Kyriss et al. Reference Kyriss, Pötschke-Langer and Grüning2008; Lopipero et al. Reference Lopipero, Apollonio and Bero2008) to have negative effects on the adoption of tobacco control policies. Furthermore, the literature has emphasised the positive effect of federalism on the adoption of tobacco control policies. Notably, the autonomy of subnational governments provides a policy laboratory for innovative tobacco control policies (Studlar Reference Studlar2002, Reference Studlar2010; Marmor and Liebermann Reference Marmor and Liebermann2004; Studlar et al. Reference Studlar, Christensen and Sistari2011, 738), and multiple venues at the federal level create opportunities for advocates of anti-smoking policies (Albaek et al. Reference Albaek, Green-Pedersen and Nielsen2007, 2). Concerning Switzerland, research on tobacco control has focussed on voters’ approval of a specific policy, either in a national context (Cornuz et al. Reference Cornuz, Burnand, Kawachi, Gutzwiller and Paccaud1996) or in the context of a specific municipality or setting (Marques-Vidal et al. Reference Marques-Vidal, Melich-Cerveira, Paccaud, Waeber, Vollenweider and Cornuz2010; Ritter and Elger Reference Ritter and Elger2014). Others have researched the historical dimension of tobacco control (Cranmer et al. Reference Cranmer, Petit and Sager2011) or the strength of the tobacco industry (Lee and Glantz Reference Lee and Glantz2001). Despite the aforementioned efforts, there is still a lack of empirically founded articles on tobacco control in Switzerland that speak to the political science literature.
A significant share of the tobacco control policy literature has focussed on diffusion effects. Broadly defined, diffusion refers to the interdependence of states and/or regions and its effect on policy adoption (Walker Reference Walker1969; Berry and Berry Reference Berry and Berry1990; Simmons and Elkins Reference Simmons and Elkins2004; Braun and Gilardi Reference Braun and Gilardi2006; Simmons et al. Reference Simmons, Dobbin and Garrett2006; Dobbin et al. Reference Dobbin, Simmons and Garrett2007). From a general perspective, authors have distinguished four forms of policy diffusion. These are coercion by soft or hard power (Simmons et al. Reference Simmons, Dobbin and Garrett2006, 790–791), competition of different policy models such as tax rates (Simmons et al. Reference Simmons, Dobbin and Garrett2006, 792–795), learning from others’ experiences (Gilardi et al. Reference Gilardi, Füglister and Luyet2009; Gilardi Reference Gilardi2010) and emulation of policies from other countries because they adhere to shared norms (Braun and Gilardi Reference Braun and Gilardi2006, 311–312). A magnitude of empirical analyses has verified these forms of diffusion empirically.Footnote 2 Related concepts such as policy transfer and lesson drawing (Dolowitz and Marsh Reference Dolowitz and Marsh1996, Reference Dolowitz and Marsh2012; Benson and Jordan Reference Benson and Jordan2011) point to similar problems. The difference between policy transfer and diffusion is that the latter focuses strictly on the process, for example, how one government learns from another, whereas the policy transfer literature also takes into account the content of the transferred policy (Dolowitz and Marsh Reference Dolowitz and Marsh1996, 344–345). Researchers have demonstrated that diffusion processes are important to explain policy differences between Swiss cantons. For example, cantonal governments learn from each other’s health insurance policies (Gilardi and Füglister Reference Gilardi and Füglister2008; Füglister Reference Füglister2012) and are less competitive in taxation towards culturally similar cantons (Gilardi and Wasserfallen Reference Gilardi and Wasserfallen2016).
Diffusion effects have been important to the adoption of tobacco control policies (Studlar Reference Studlar1999; Brandt Reference Brandt2004). According to the literature, learning (Shipan and Reference Shipan and VoldenVolden Reference Shipan and Volden2014) plays a big role in anti-smoking policies, but emulation, competition and coercion are also highly relevant factors (Shipan and Reference Shipan and VoldenVolden Reference Shipan and Volden2006, Reference Shipan and Volden2008). Apart from the research focussing on the United States (US), the authors point to diffusion effects in tobacco control policy relating to Canada, Australia and New Zealand (Studlar Reference Studlar2005), as well as in Europe. The research looking at tobacco control policy in Europe underlines that the EU had an important impact on the adoption of tobacco control policies (Duina and Kurzer Reference Duina and Kurzer2004; Studlar Reference Studlar2006a; Asare et al. Reference Asare, Cairney and Studlar2009). In particular, authors have argued that EU action in the field of tobacco control policy has learning effects on its member states regarding anti-smoking policy (Duina and Kurzer Reference Duina and Kurzer2004; Studlar Reference Studlar2006a, 386–387).Footnote 3
More recent contributions have underlined the importance of a positive public opinion to the adoption of new tobacco control policies (Toshkov Reference Toshkov2013). Pacheco (Reference Pacheco2012, 187) has introduced the social contagion model into the research on policy diffusion. Her main argument is that voters’ approval of tobacco control puts the issue on policymakers’ agendas. In other words, the adoption of a policy in state A leads to public support for the policy in state B, which puts the policy idea on decisionmakers’ agendas in state B (Pacheco Reference Pacheco2012, 189, Reference Pacheco2011, Reference Pacheco2013). This contribution lends itself to the literature on attitudinal policy feedback that is interested in the connection between policies and mass attitudes in all kinds of policy fields, such as abortion, healthcare or welfare (Hill et al. Reference Hill, Leighley and Hinton-Andersson1995; Gray et al. Reference Gray, Lowery and Godwin2007; Camobreco and Barnello Reference Camobreco and Barnello2008). One general finding of this literature is that citizens learn about policies, and consequently support or refuse further interventions in a given policy area. The drivers of support might be objective conditions (Johnson et al. Reference Johnson, Brace and Arceneaux2005), subjective perceptions caused by elite framing (Hetling and McDermott Reference Hetling and McDermott2008) or self-interests (Schneider and Jacoby Reference Schneider and Jacoby2007). In particular, voters’ support for tangible policies ought to be influenced by their experiences: bad experiences with a given policy will erode mass support; good ones increase it (Pacheco Reference Pacheco2013).Footnote 4
To summarise, for the purposes of this article, three points are important regarding diffusion, tobacco control and mass attitudes on policies. First, tobacco control policies spread by diffusion, and there is demand for more research on tobacco control policy in Switzerland. Second, the EU played an important, but not exclusive, role in its adoption in Europe. Finally, mass attitudes are important for policy change in general, but also regarding anti-smoking policies. Below, the article uses these insights to develop a new argument about the connection between attitudes towards the EU and adoption of advertising regulations in Switzerland.
Theoretical framework
This article’s main argument forges a link between popular support for European integration and policy adoption – namely, the restriction of tobacco advertising in Swiss cantons. Precisely, the article argues that higher support by voters for European integration increases the probability of banning advertisements for tobacco.
Support for more European integration
To develop its key argument, this article refers to the “social contagion model” in policy diffusion research (Pacheco Reference Pacheco2012, 189–190). According to Pacheco, the adoption of a policy in state A influences public opinion concerning the policy in state B. Through direct experience, social networks or the media, citizens in state B may learn about, change their opinion on and voice approval of the policy. Consequently, policymakers in state B implement the policy in their territory because they fear no voter opposition (Pacheco Reference Pacheco2012, 189). Thus, citizens learn from policies – be it through changing objective conditions (Johnson et al. Reference Johnson, Brace and Arceneaux2005), subjective perceptions driven by elite framing (Hetling and McDermott Reference Hetling and McDermott2008) or self-interests (Schneider and Jacoby Reference Schneider and Jacoby2007) – and policymakers account for public opinion when deciding on the reform.
This article’s argument starts with insights from Pacheco’s social contagion model; however, it develops the model a small step further by focussing on voters’ opinions about the state, or country, a policy originates from – next to the electorate’s preferences about a specific policy. Researchers working on topics such as economic voting and electoral politics have argued that voters use cognitive shortcuts such as the situation of the national economy (Duch and Stevenson Reference Duch and Stevenson2008) or party identification to decide whether they support certain policies, political parties or presidential candidates in specific elections (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Jacoby Reference Jacoby1994; Nadeau and Lewis‐Beck Reference Nadeau and Lewis‐Beck2001; Stevenson Reference Stevenson2001). Building on these insights from the literature on economic voting and electoral politics, this article argues that voters use their preferences about countries, regions or jurisdictions, which serve as a model for a certain policy, as cognitive shortcuts to judge whether they should support the policy or not. Such logic can affect policymakers’ decisions to adopt a certain policy, especially a contentious one like tobacco advertisement restrictions. If there is no particular opposition to the country from where a policy originates, policymakers are more likely to adopt it. Nevertheless, if opponents of a policy can mobilise hidden resentments against the country in which a policy originates, chances that the government will adopt it decrease, especially if the reform proposal might become subject to a popular vote. This argument complements the social contagion model, as it emphasises the relationship between the introduction of a policy in another jurisdiction and the support for that policy at home, which is rather underspecified in the original model.
Such an assumption is plausible to make the case of tobacco advertisement bans in Switzerland, for a number of reasons. First, the literature has shown that diffusion effects are important between Swiss cantons (Gilardi and Füglister Reference Gilardi and Füglister2008; Füglister Reference Füglister2012). Second, Europe and Switzerland are multicultural contexts with different languages and nationalities (Lacey Reference Lacey2014), and voters have preferences about other cantons, countries or the EU. Approval of a policy might not only be because of its substance and personal experience but also its place of origin. If factual information is lacking, voters and policymakers look for signals for how to discriminate. For example, voters use the state of the economy (Nadeau and Lewis‐Beck Reference Nadeau and Lewis‐Beck2001; Duch and Stevenson Reference Duch and Stevenson2008) or their identification with a given party (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Jacoby Reference Jacoby1994; Stevenson Reference Stevenson2001) as signals to discriminate whether they support that party or candidate in an election. Similarly, policymakers use macroeconomic policies in a way that pleases voters, knowing that the electorate mobilises the state of the economy as a signal to judge the performance of the incumbent (Nordhaus Reference Nordhaus1975). In Switzerland, support for more European integration is a telling shortcut for voters and policymakers, because it is a highly contentious issue. After voters refused to enter the EEA in 1992 by a tiny majority, Switzerland closed a number of bilateral treaties (Kriesi and Trechsel Reference Kriesi and Trechsel2008, 176–186). Voters approved these treaties, but French-speaking cantons supported these treaties more than German-speaking ones.Footnote 5
Third, Switzerland has discussed tobacco advertising restrictions as a European issue. For example, the high-quality newspaper Neue Zürcher Zeitung (NZZ) continuously covered tobacco advertisement restrictions in an EU context. Between 1997 and 2010, 31.25% of the NZZ articles dealing with tobacco advertisement restrictions discussed them as a European topic and evaluated their possible negative effects on the Swiss economy.Footnote 6 Therefore, citizens and policymakers connected them to European integration.Footnote 7 Consequently, it is plausible to assume that cantonal policymakers consider the approval of European integration among the electorate before deciding to adopt a policy.
Fourth, there is a lack of research-based information that Swiss policymakers could use to convince voters, despite existing evidence that regulation of tobacco advertising reduces smoking rates (Pierce et al. Reference Pierce, Gilpin, Burns, Whalen, Rosbrook, Shopland and Johnson1991; Leeflang and Reuyl Reference Leeflang and Reuyl1995; Saffer and Chaloupka Reference Saffer and Chaloupka2000; Blecher Reference Blecher2008). The problem is that there is no sufficient data available on changes in smoking behaviour at the cantonal level (Keller et al. Reference Keller, Radtke, Krebs and Hornung2011a, Reference Keller, Radtke, Krebs and Hornung2011b). As strong data that might prove advertisement bans’ success (Gilardi et al. Reference Gilardi, Füglister and Luyet2009) are lacking, cantonal governments need to base their case for such bans on normative grounds. In particular, they need to convince other parties and interest groups – as well as the population – that such a policy, which has a clearly Europeanised component, should also be adopted in their canton.
To sum up, there are strong reasons to argue that if voters approved of deeper European integration, policymakers would use this as a signal that voters would also support restrictions on tobacco advertising. Thus, the governments of cantons with higher support for European integration are more likely to ban tobacco advertising because they do not fear punishment from the electorate if they adopt ideas connected to Brussels.
Consequently, the article puts forward the following hypothesis: The higher the popular support for European integration, the more likely cantonal governments are to adopt tobacco advertisement restrictions.
Controls
Nevertheless, there are a number of other factors that could explain why some cantons restricted tobacco advertisements while others did not. In the following sections, the article presents these elements. Thereby, it puts a special emphasis on two of them – tobacco in the cantonal economy and coordination in the board of Cantonal Directors of Public Health – as they refer to different diffusion mechanisms.
Tobacco in the cantonal economy
Even if the support for more European integration is comparatively strong, policymakers might decide not to ban tobacco advertising, simply because the tobacco industry or tobacco cultivation is relevant to a canton’s economy. In the diffusion literature, authors refer to economic competition (Braun and Gilardi Reference Braun and Gilardi2006, 308) as a reason why governments do not imitate policies of others. This can be for tax competition (Gilardi and Wasserfallen Reference Gilardi and Wasserfallen2016); however, in the case of tobacco control, the reason can be to keep the tobacco industry in the canton (Lee and Glantz Reference Lee and Glantz2001). This logic can also be found in the literature on tobacco control policy, which has shown that a strong tobacco industry presence restrains the introduction of tobacco control, sometimes by threatening to leave. International tobacco corporations, as well as local producers, influence policy through direct lobbying of politicians, such as parliamentarians or members of government, but also by influencing research on the negative effects of smoking, especially passive smoking (Bornhäuser et al. Reference Bornhäuser, McCarthy and Glantz2006; Studlar Reference Studlar2006b; Grüner et al. Reference Grüner, Strünck and Gilmore2008; Kyriss et al. Reference Kyriss, Pötschke-Langer and Grüning2008; Lopipero et al. Reference Lopipero, Apollonio and Bero2008).
For Switzerland, this implies that multinational tobacco corporations or tobacco cultivation should reduce the boost that support for more European integration has on the adoption of tobacco control or enforce the negative impact of low support for deeper integration. With the tobacco industry present, cantonal governments are afraid that restrictions on tobacco advertising might drive tobacco corporations or farmers to other cantons, harming their canton’s economy. Thus, the presence of the tobacco industry or tobacco farming in the canton reduces the probability of adopting tobacco advertisement restrictions.
Membership in intergovernmental organisations: learning in the board of the CDP (Conference of Cantonal Directors of Public Health).
Another control variable accounts for policy coordination between cantonal governments: Notably, interaction in the board of the CDP could explain why some cantons are more likely to adopt ad restrictions than others. Swiss cantonal policymaking is sometimes explicitly or implicitly coordinated. In health policy, the agency of such coordination is the CDP. The CDP is part of the highly institutionalised system of intercantonal conferences, which also deals with other policy areas such as economy, finances or education (Bochsler and Sciarini Reference Bochsler and Sciarini2006; Bolleyer Reference Bolleyer2009). These conferences increase the capacity to coordinate policies, but are not obliged to find binding solutions, which means that cantons might unilaterally decide to pursue different policies (Bolleyer Reference Bolleyer2006). The goal of the CDP is to promote the coordination and cooperation of all 26 cantons regarding health policy, along with the federal government and the principality of Liechtenstein. It is primarily concerned with healthcare policies such as the building of hospitals and coordination of health insurance premiums; however, if necessary, it also coordinates preventive health policies. A board, elected by the plenary session of the cantonal members, organises and leads the CDP.Footnote 8
In the conference, cantonal delegates do not represent a specific position, but are free to negotiate compromises and exchange information. As a consequence, there is room for mutual exchanges with regard to health policies, especially among board members. A recent contribution by Füglister shows that cantons that actively participate on the board of the CDP learn from one another and imitate each other’s successful policies (Füglister Reference Füglister2012). Therefore, if a cantonal minister of health is a member of the CDP board, his or her canton is more likely to regulate tobacco advertising. This is plausible because the CDP’s annual reports show that both the general assembly and the board discuss tobacco control, including advertisement restrictions. In 2003, the board recommended that all cantons restrict tobacco advertising.Footnote 9 In interviews, the former secretary of the CDP and the former head of the Federal Office for Public Health confirmed formal and informal exchanges on tobacco control policy in this arena.Footnote 10
Further explanations
There are further explanations for differences in the adoption of tobacco advertising regulations between Swiss cantons, for which an empirical analysis must control. These are either elements outlined as important by the tobacco control literature, such as public support for tobacco advertising restrictions (Pacheco Reference Pacheco2012; Toshkov Reference Toshkov2013), previous adoption in neighbouring “states” (Shipan and Reference Shipan and VoldenVolden Reference Shipan and Volden2006) or a high percentage of the left party (Duina and Kurzer Reference Duina and Kurzer2004; Cairney Reference Cairney2007) in the cantonal government. All these elements should affect tobacco control. The overall size of the cantonal population and the German-speaking percentage (Studlar et al. Reference Studlar, Christensen and Sistari2011) are possible confounders as well. There are no clear expectations concerning language; however, larger cantons should implement tobacco advertisement restrictions faster because their administrative capacity is bigger (Vatter Reference Vatter2014b, 252–253). Harkening back to research on policy diffusion in Switzerland, the article controls for other elements: proportion of the Liberal Party, Christian Democrats and Swiss People’s Party in cantonal governments (Gilardi and Füglister Reference Gilardi and Füglister2008; Gilardi and Wasserfallen Reference Gilardi and Wasserfallen2016) and cultural proximity (Gilardi and Wasserfallen Reference Gilardi and Wasserfallen2016). Although there is no clear expectation regarding the positions of the mentioned political parties concerning tobacco advertisement bans, cultural closeness makes adoption more likely. Furthermore, the cantons differ sharply regarding their regulatory density, that is, regulations in other fields of civil life, such as schooling, church taxes and alcohol consumption (Hämmerli and Rutz Reference Hämmerli and Rutz2014), and the analysis must control for this factor. Thereby, a higher number of civil life regulations should increase the likelihood of adopting tobacco advertising restrictions.
EU relations and tobacco control policy in Switzerland
Next to the fact that there is still a demand for research on tobacco control policy in Switzerland, the country is an interesting example of the indirect impact of the EU on its neighbouring countries. Switzerland is a member of the European Free Trade Area along with Iceland, Liechtenstein and Norway. Unlike those three countries,Footnote 11 Swiss voters refused to enter the EEA in a popular vote in 1992. Therefore, Switzerland negotiated a number of bilateral treaties with the EU, according to which it applies some EU regulations in order to have full access to the common market (Kriesi and Trechsel Reference Kriesi and Trechsel2008, 177–183).
In contrast to other policy fields, Swiss tobacco control followed only indirectly the general Europeanisation of Swiss politics (Mach et al. Reference Mach, Häusermann and Papadopoulos2003; Sciarini et al. Reference Sciarini, Fischer and Nicolet2004; Fontana Reference Fontana2011). Since 2005, some of the instruments developed regarding Swiss tobacco control (Joossens and Raw Reference Joossens and Raw2007, Reference Joossens and Raw2011, Reference Joossens and Raw2014) resemble recommendations of the EU – for instance, increased spending on anti-smoking campaigns and smoking bans (Studlar Reference Studlar2006a, 387). However, the national government did not adopt the EU directive on tobacco advertising (2003/33/EC). Students of European tobacco control have considered Switzerland a laggard, but the country has improved, especially since 2004. Switzerland ranked 21 (out of 28) in 2004, 24 (out of 30) in 2005, 18 (out of 30) in 2007, 11 (out of 34) in 2010 and 18 (out of 34) in 2013. The country’s ranking improved because the national government passed a series of reforms, such as requiring warning labels on tobacco products and increasing tobacco taxes and investment in tobacco control in general, as well as a limited national smoking ban. However, there are shortcomings in the country’s tobacco control policy. Notably, the authors of the tobacco control scale justify the 2013 downgrading of Switzerland because of its lack of a national ban on tobacco advertising and failure to ratify the Framework Convention on Tobacco Control (Joossens and Raw Reference Joossens and Raw2007, Reference Joossens and Raw2011, Reference Joossens and Raw2014, 10).
In Switzerland, it is difficult to establish a comprehensive tobacco control regime because the country is decentralised and has many veto possibilities at the national level. The OECD (Organisation for Economic Co-operation and Development) pointed to this problem in its review of the Swiss health system (OECD and WHO 2006; OECD 2011), as does the public health literature (Mayes and Oliver Reference Mayes and Oliver2012). So far, the existing contributions on tobacco control policy only hint to possible explanations, such as the strength and political influence of the nation’s tobacco industry (Lee and Glantz Reference Lee and Glantz2001). On the other hand, direct democracy allows for presenting popular initiatives with innovative policies. This happened twice concerning tobacco advertising. In 1993, a popular initiative proposing to ban all tobacco and alcohol advertising went up for a vote. A large majority of voters rejected it. Clearly, the tobacco industry had influenced the campaign, but the high degree of popular refusal caused federal policymakers to shy away from tobacco control policy for some time (Cornuz et al. Reference Cornuz, Burnand, Kawachi, Gutzwiller and Paccaud1996).
On the other hand, Switzerland is a very decentralised federation of 26 “member states” (cantons), which are quite different culturally and politically (Kriesi and Trechsel Reference Kriesi and Trechsel2008; Vatter Reference Vatter2014a). Consequently, it is also a policy laboratory because cantonal governments can test innovative tobacco control policies, as some existing research on tobacco control suggested it for other countries (Studlar Reference Studlar2002, Reference Studlar2010). Nevertheless, because of its strong consensus culture, cantonal governments are regularly involved in the consultation procedure of the national parliament, which sets federal policy (Linder Reference Linder1994; Kriesi and Trechsel Reference Kriesi and Trechsel2008) and interacts horizontally regarding policies and their implementation (Bochsler and Sciarini Reference Bochsler and Sciarini2006; Bolleyer Reference Bolleyer2009). Although there is little formal coordination of policy outputs between all cantons, there is a lot of exchange of information and best practices. Consequently, diffusion mechanisms such as competition and learning remain important (Füglister Reference Füglister2012; Gilardi and Wasserfallen Reference Gilardi and Wasserfallen2016).
Concerning tobacco control, competences are separated and shared between the federal government and the cantons. In most domains, the federal government is responsible, whereas the cantons have leverage of their own for some instruments in case there is no overriding federal law. The federal government is in charge of prices and tax measures, second-hand smoke issues, tobacco product disclosure, packaging, public information campaigns, national advertising and international collaboration. Cantonal responsibilities include second-hand smoke protections, public information, tobacco advertising and sponsorship, demand reduction and protection of minors. The cantons passed legislation, especially about tobacco advertising, smoking bans and youth protection; however, in the latter two fields, the federal government has harmonised existing legislation to some extent. In 2010, a national smoking ban was passed; currently, legislation concerning sales restrictions and national law on tobacco is in development.Footnote 12 In addition, many cantons have comprehensive tobacco control strategies.Footnote 13
Cantons and the federal government share legislative responsibility for tobacco advertisement restrictions. The latter is responsible for tobacco advertisement bans on television and radio. Switzerland was, in 1964, one of the first countries to ban tobacco advertising on television and the radio.Footnote 14 After that, however, federal tobacco advertising restrictions have been difficult. In 1979 and 1993, two popular initiatives attempted to introduce national smoking bans in Switzerland, but voters rejected both (Cornuz et al. Reference Cornuz, Burnand, Kawachi, Gutzwiller and Paccaud1996). In 1995, a decree by the federal government banned tobacco advertisements in all forms of media directed towards young people.Footnote 15 Yet, further legislative initiatives regarding advertising restrictions at the national level failed.
Unlike other EU policies (Mach et al. Reference Mach, Häusermann and Papadopoulos2003; Sciarini et al. Reference Sciarini, Fischer and Nicolet2004; Fontana Reference Fontana2011), the national government in Switzerland did not follow the European initiative regarding tobacco advertising. This was because of politicisation of the topic (Cornuz et al. Reference Cornuz, Burnand, Kawachi, Gutzwiller and Paccaud1996); as tobacco control became part of the bilateral agreement concerning health between the EU and Switzerland, negotiations over this are currently suspended.Footnote 16 However, some cantonal governments took the initiative and adopted tobacco advertisement restrictions. In 2000, the canton of Geneva banned all tobacco and alcohol advertisements visible from public grounds. In 2002, the federal court of Switzerland declared the law in Geneva to be constitutional. It held that the responsibilities of the federal government are only perfunctory in this policy area; therefore, the cantons may implement their own tobacco ad restrictions.Footnote 17
After this ruling, more cantons passed tobacco ad restrictions. Figure 1 shows the number of cantons per year that implemented restrictions on tobacco advertising between 1998 and 2010.Footnote 18 As you can see, Geneva was the forerunner, and other cantons implemented bans mostly after 2005. The last cantonal restriction at this writing came into force in 2010 (Zug). In 2015, 15 cantons implemented tobacco advertisement restrictions.Footnote 19 Nevertheless, there has so far been no research to explain these differences between cantons.
Adoption of tobacco ad bans in the Swiss cantons.
To sum up, analysing tobacco ad restrictions at the level of Swiss cantons is interesting for two reasons. First, the presence of 26 culturally and politically different cantons provides an interesting laboratory for researching the indirect impact of the EU on its neighbours. Second, such analysis boosts our understanding of tobacco control, policy diffusion and the Europeanisation of public policies in Switzerland.
Empirical analysis
This article uses EHA to analyse the above-made argument empirically. Below, the article discusses operationalisation of variables and choice of method. After that, the results of the empirical analysis will be presented.
Data and methodFootnote 20
The analysis uses a time series cross-section data set. Data collection began in 1998, when the first EU Directive on tobacco control policy was passed, and ended in 2010, when the last canton adopted a tobacco advertisement regulation.Footnote 21 The dependent variable is binary: 0 by default and 1 once a tobacco advertisement restriction has been adopted.Footnote 22
This article operationalises the independent variables as follows. To measure support for more European integration, the article uses the results of the popular votes on the relationship of Switzerland with the EU.Footnote 23 The higher the support for these treaties, the more likely it is for cantonal governments to restrict or ban tobacco advertising. The tobacco industry takes the value of 1 if the canton is home to a multinational tobacco industry and/or has tobacco planters. In this case, the article expects a negative effect on tobacco control.Footnote 24 For previous policy adoption in a neighbouring canton or country, the article uses two different variables with different measurements. First, it uses a continuous variable that simply counts the neighbouring countries or cantons with a tobacco ad ban. The more neighbouring cantons or countries that banned tobacco advertising, the more likely becomes a ban in a given canton. Second, it uses a spatial lag based on a connectivity matrix of cantonal neighbourhoods, which accounts for neighbouring cantons but not countries. For example, unlike in the first variable, the fact that the canton of Geneva has borders to the canton Vaud is taken into consideration but not its proximity to France. Thus, the expectations are the same.Footnote 25 Concerning membership on the board of the CDP, the article counts the years a canton is a member of the board. This should also promote the banning of tobacco advertising. The article operationalises the strength of the left party in government using the percentage of seats in the cantonal governments occupied by the Social Democrats and Greens [find a similar procedure in Freitag and Vatter (Reference Freitag and Vatter2008) and Füglister (Reference Füglister2012)]. To account for the share of the Swiss People’s Party, the Christian Democrats and the Liberals, the article uses the same measure. The measure of public opinion is a snapshot of the results of the 1992 referendum on tobacco advertisement bans in Switzerland.Footnote 26 Higher percentages in these two variables should increase the canton’s likelihood of banning tobacco ads. To account for the regulation of other civil liberties, the article uses an indicator to measure regulation in other fields of civil life. The higher the score on this indicator, the more likely the adoption of advertisement restrictions by the cantonal government.Footnote 27
As a method, the article uses EHA based on logit models. This has become a standard in political science, especially regarding the analysis of policy diffusion (Berry and Berry Reference Berry and Berry1990; Shipan and Reference Shipan and VoldenVolden Reference Shipan and Volden2008; Gilardi et al. Reference Gilardi, Füglister and Luyet2009; de Francesco Reference de Francesco2012; Füglister Reference Füglister2012; Shipan and Reference Shipan and VoldenVolden Reference Shipan and Volden2014; Prorok and Huth Reference Prorok and Huth2015). In EHA, it is important to set a meaningful period for analysis. The data set starts in 1998 because the EU adopted its first tobacco advertisement regulation that year. When using time series cross-section data, it is necessary to control for time; otherwise, the model assumes that all observations are independent of each other. To account for this problem, the article includes three time variables for estimating the discrete EHA – t, t 2/10, t 3/100 – which provides a more accurate control for time (Buckley and Westerland Reference Buckley and Westerland2004; Carter and Signorino Reference Carter and Signorino2010) than simple time dummies (Beck et al. Reference Beck, Katz and Tucker1998). Political science literature widely uses this practice (de Francesco Reference de Francesco2012; Prorok and Huth Reference Prorok and Huth2015).
Results
Table 1 shows the results of the discrete EHA for the adoption of tobacco advertisement restrictions in 26 Swiss cantons. Standard errors are robust (clustered by canton), the results do not change in alternative specifications and the results pass a test for omitted variable bias.Footnote 28 To put the discussed hypothesis to an empirical test, the article refers to seven models. The first one includes support for more European integration and the controls for time. The second model adds the dimension of economic competition, whereas Model 3 adjusts for the time a canton has spent on the board of the CDP. Models 4 through 6 include the mentioned additional controls. The last model controls for the spatial lag variable instead of the continuous variable that entails the number of neighbouring countries or cantons.
Discrete event-history analysis (robust standard errors – clustered by canton)
Note: Standard errors in parentheses.
Significance levels: *p<0.1, **p<0.05, ***p<0.01.
The regression coefficients point in the directions expected after the theoretical discussion. Support for more European integration correlates with the adoption of tobacco advertisement bans. However, if estimated with the variable for European integration support only, the coefficient is not significant statistically. Statistical significance can only be attained when controlling for the presence of a tobacco multinational or if tobacco is cultivated in the canton (Models 2–7). Hence, it is essential to consider support for European integration along with tobacco industry presence or cultivation in the canton. This finding supports the argument that policymakers use voters’ approval of the EU as a signal for the support of tobacco advertising restrictions. Nevertheless, the result also suggests that policymakers are especially likely to account for the public’s support of European integration in case the tobacco industry or tobacco cultivation is not present in the canton. Put differently, the effect of the European dimension is particularly clear once there are no adverse special interests in place.
Participation on the board of the CDP increases the likelihood of tobacco advertising regulation, but the results are not statistically significant, except for Model 3. Concerning the control variables, the results show positive effects for public opinion on tobacco advertising restrictions, previous adoption in a neighbouring canton or country and the percentage of left parties in the government. The fact that the public opinion variable is significant supports Pacheco’s findings and shows that it is relevant for Switzerland as well.
Except for left parties, the coefficients are statistically significant. The domestic spatial lag variable accounts for the number of neighbouring cantons with a ban but does not include common borders of cantons with EU countries, which might also incentivise the government of a canton bordering an EU country to restrict tobacco advertising. This variable has a negative and statistically insignificant regression coefficient and suggests that domestic diffusion does not explain the variance between cantons regarding tobacco advertisement restrictions. However, if we replace the domestic spatial lag with a variable that adds the presence of a tobacco advertisement restriction in a neighbouring EU country to the presence of a ban in neighbouring cantons, the diffusion variable shows a positive effect. This finding implies that cantons bordering EU countries and cantons that restrict tobacco advertising are more likely to restrict tobacco advertising themselves than those cantons that have no common borders with a tobacco-banning EU country but share frontiers only with another canton that has already banned tobacco advertising. Such a finding points at another pathway according to which the European dimension influences the adoption of tobacco advertisement restrictions: common borders with an EU country that has already restricted tobacco advertising makes cantonal adoption more likely. The diffusion literature previously pointed to this channel of diffusion (Maggetti and Gilardi Reference Maggetti and Gilardi2016), and according to this analysis it matters for the Swiss case, in addition to public support for the EU.
The size of the German-speaking population and cantonal population overall have no clear effect. Concerning the additional partisan controls, a higher percentage of the Swiss People’s Party and the Christian Democrats has a particularly positive effect, whereas more seats for the Liberal Party have a negative effect. This suggests that tobacco advertising restrictions in Switzerland are an economic issue because liberal parties have had a positive record of tobacco control policy in other countries (Studlar Reference Studlar2005; Hooker and Chapman Reference Hooker and Chapman2006). Two of the three time variables are negative and highly significant in all models. This finding suggests that the likelihood of adoption decreases as more time passes since tobacco advertising restrictions entered the political agenda. In other words, any given canton is most likely to ban tobacco advertisement soon after 1998 – everything else constant – and such action becomes less likely as more years pass.
To understand better the results concerning the effect of the European dimension, we turn to a graphical interpretation. Figure 2 shows the predictive margins for the restriction of tobacco advertisement across the magnitude of support for the European integration in popular votes, one with the tobacco industry present and one with it absent. The picture demonstrates clearly that the likelihood of restricting tobacco advertising increases with support for further European integration. However, the graph also shows that this effect is especially strong if there is neither a multinational tobacco corporation nor tobacco cultivation in the canton. In terms of overall marginal effects, the results show that an increase in popular support for more European integration by 1% augments the likelihood for the canton to adopt a tobacco advertisement ban by 0.54%. The change from no tobacco industry or tobacco cultivation in the canton to the presence of one or the other decreases the likelihood of an advertising ban by 82%.Footnote 29 The effects for the tobacco industry are particularly strong, but the coefficients are not statistically significant. The conditional marginal effects in Figure 2 show that the probability to adopt tobacco advertising restrictions increases by 20% if the support for European integration increases from 60–80% of voters and if the tobacco industry is not present in the canton. In case tobacco manufacturing and cultivation happen in the canton, higher support for the EU also increases the likelihood to restrict tobacco advertisement; however, the estimator only shows a statistically significant increase in the probability to adopt a tobacco advertisement ban if more than 70% of the cantonal voters support European integration.
Predictive margins for European integration support and tobacco industry.
Overall, these results support the above-raised hypotheses that higher support for deeper European integration makes it more likely that cantonal governments will implement tobacco advertising restrictions because they have popular support for these regulative measures. Furthermore, the results suggest that policymakers are particularly inclined to account for EU approval if adverse special interests – that is, the tobacco industry and growers – are absent.
A different explanation for the presence of tobacco advertisement restrictions the article has put forward is the number of years a cantonal government participated on the board of the CDP. Figure 3 presents the predictive margins of participation years on the board of the CDP, conditional on different configurations of EU support and the presence of the tobacco industry. The results show that – with the tobacco industry present – longer participation on the CDP board makes adoption of tobacco advertising restrictions more likely when popular support for European integration is strong. If voters’ support for European integration is weak, the effect of the cantonal governments’ participation on the CDP board on adopting restrictions on tobacco advertising is also feeble – in fact barely visible compared with when EU support is strong. Specifically, Figure 3 suggests that the conditional effects of membership on the board of the CDP on banning tobacco ads increase approximately by 10% between 0 and 13 years of participation when voters support European integration. If voters do not support European integration, the probability to restrict tobacco advertising increases by 0.1% between 0 and 13 years of participation in the CDP.Footnote 30 This result indirectly supports the argument put forward in the article, as it underlines that cantonal governments tended to coordinate their policies on restricting tobacco advertising only if voters supported European integration.
Predictive margins for time on the CDP board.
These findings are important to the research concerning Switzerland in at least two ways. First, the article analysed an important dimension of tobacco control policymaking, the adoption of advertisement restrictions in the cantons. This research shows that some cantons ban advertisements of tobacco products to a certain extent, whereas others flinch from such measures. The analysis demonstrates that tobacco advertisement restrictions strongly correlate with support for more European integration and the presence of the tobacco industry. Second, this is an important finding for research on the relationship between the EU and Switzerland. Specifically, this article implies that decisions and political processes in Brussels extend not only to the federal level (Mach et al. Reference Mach, Häusermann and Papadopoulos2003; Sciarini et al. Reference Sciarini, Fischer and Nicolet2004; Fontana Reference Fontana2011) but also to the cantonal level in Switzerland. According to this article’s results, high public support for more European integration, favourable economic conditions and coordination between the cantonal governments are important elements for the subnational implementation of policies that resemble European directives. This hypothesis could be tested in further research on subnational policymaking in Switzerland, such as immigration policy (Manatschal Reference Manatschal2011).
Relevance beyond the Swiss case
Although the empirical analysis in this article is limited to tobacco advertising in Switzerland, this article allows researchers to draw some conclusions of general interest to political science in two ways. First, the article contributes to the literature on policy diffusion (Walker Reference Walker1969; Braun and Gilardi Reference Braun and Gilardi2006; Dobbin et al. Reference Dobbin, Simmons and Garrett2007; Shipan and Reference Shipan and VoldenVolden Reference Shipan and Volden2008; Gilardi Reference Gilardi2013; Maggetti and Gilardi Reference Maggetti and Gilardi2016). A recent contribution to this literature is the “social contagion model” (Pacheco Reference Pacheco2012), which assumes that the opinions of residents of a country or state about a policy determine whether the government adopts it. The article has used this model to argue that, next to preferences about the policy itself, voter opinion of its country of origin matters. Public approval of a political entity affects whether governments emulate its policies. For example, a policy originating in the US might first be adopted in countries where the US has a good reputation rather than in countries where its reputation is bad, independent of its success. It would be interesting to test such a hypothesis using a larger sample of countries and a set of different policies. Furthermore, researchers could analyse whether democratic countries adopt policy instruments designed by autocratic governments, for example, in the field of environmental policy. The operationalisation of the emulation dimension, which is linked to norms, could especially benefit from such data because it entails more information about the agent of a norm than the number of previous adopters.
Second, the article shows that the EU’s role as a regional political and social model (Schimmelfennig Reference Schimmelfennig2009) affects the subnational level of neighbouring countries. This might cause conflicts between regions and the national government in federal states (Braun and Trein Reference Braun and Trein2014) such as in Switzerland, but also in unitary states such as Norway, where regions could use their autonomy to orient their policies towards the EU, depending on the population’s approval. The idea of subnational governments as policy laboratories is not new (Pierson Reference Pierson1995, 470). However, in combination with the EU, this might lead to new conflicts between national and subnational governments because the latter are more progressive in implementing EU guidelines and directives. The implications are different for weak states on the European periphery, such as Ukraine. According to this research, a country’s provinces may differ in approval of more European integration. In Ukraine’s western provinces, the population should be much more favourable towards the EU than in the eastern part of the country. Such differences in public opinion fuel continuing conflict between the parties.
Conclusions
This article started with the empirical problem that the EU banned tobacco advertising and that some of the Swiss cantons restricted tobacco advertisement, but others did not. The article has shown that, among others, support for more European integration by voters encouraged cantonal governments to restrict tobacco advertising. This article has argued that citizens’ approval of deeper European integration influences whether a cantonal government bans tobacco advertisement. In particular, if citizens favour European integration, cantonal governments are more likely to adopt advertising restrictions the EU previously adopted. This article’s empirical analysis has verified this hypothesis using statistical evidence, and has shown that support for further European integration strongly correlates with the adoption of advertising restrictions, especially if the economic situation in the canton is favourable to regulation.
Yet, there are some limitations to the analysis this article has put forward. First, the empirical results are based on statistical analyses only. Although the article has found ample evidence that tobacco advertising restrictions were discussed as a Europeanised issue in Switzerland, the article did not conduct case studies to more deeply examine the question of how decisionmakers took approval ratings of European integration by voters into account when deciding upon restrictions of tobacco advertising. This question remains open for further research.
Nonetheless, this article contributes to the political science literature in various ways. First, the article adds to the comparative Swiss public policy literature with an analysis of the diffusion of tobacco control policies. Second, the article contributes to tobacco control policy research by putting forward an analysis of anti-smoking policies in Switzerland, which is an interesting case but for which a comprehensive analysis of tobacco control policy is still lacking. Third, the article contributes to the Europeanisation literature by pointing to the indirect effects of EU policies and directives for policies on subnational levels of government – even in countries that are not formal EU members. In addition, this research points to the question of whether deeper European integration affects the adoption culture in a country and coincides with “worlds of compliance” (Falkner et al. Reference Falkner, Treib and Holzleithner2008) and the transposition of EU soft law. Fourth, the results of this article make a small contribution to the policy diffusion literature in general. The findings in this article demonstrate that we need to complement the “social contagion model” (Pacheco Reference Pacheco2012) because – next to voters’ preferences about a given policy – the electorate’s opinion about the country or jurisdiction in which a policy originates matters for policymakers who decide on the adoption of that policy.
Acknowledgements
The author thanks Dietmar Braun, Karin Ingold, Martino Maggetti, William Ossipow, Eva Thomann, Fabio Wasserfallen, the Editors and three anonymous reviewers for excellent comments and suggestions. In addition, the authors acknowledge financial support by the Swiss National Science Foundation (Grant No. 26041044).
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0143814X16000167
