Does Participation in Policymaking Enhance Satisfaction with the Policy Outcome? Evidence from Switzerland

Swiss Political Science Review 19(1): 60 83 doi:10.1111/spsr.12021 Does Participation in Policymaking Enhance Satisfaction with the Policy Outcome? Evidence from Switzerland DENISE TRABER University of Zurich Abstract: This article investigates the benefit of participation in policymaking. Based on data from interviews with various interest groups, private organisations, and representatives of regions that had been involved in recent policymaking processes in Switzerland, this study analyses the relationship between participation and satisfaction with the policy outcome. The study looks at two different aspects of outcome satisfaction: perceived consensus of a decision and preference attainment. The results show that the level of participation is related to both aspects of outcome satisfaction but that the strength of association is dependent on the type of interest group and the level of conflict among the groups participating in the decision-making process. KEYWORDS: Decision-making, Interest groups, Participation, Public policy, Public administration Introduction It is generally acknowledged that participation is one of the most important elements of democratic systems. Any transformation of non-democratic or semi-democratic regimes involves the claim for more participation opportunities for citizens and groups representing different sections of society. Participation is said to improve responsiveness and legitimacy but also the stability of a political system. The value of participation in decision-making procedures has been of great interest in numerous research traditions. Researchers generally conclude that participation has a positive effect on the participants satisfaction with political outcomes and the legitimacy of procedure (Frey and Stutzer 2005; Gangl 2003; Hibbing and Theiss-Morse 2001; Lind and Tyler 1988; Olken 2010). When studying the effects of participation, different research traditions not only focused on different aspects of decision-making but also (sometimes implicitly) assumed different underlying causal mechanisms. Social-psychological and survey-based accounts are generally concerned with individual perception of fairness of procedure and outcome. Their The data used in this study has been collected in the research project The Swiss Decision-making System in the 21. Century: Institutions, Power, Conflict funded by the Swiss National Science Foundation (SNSF; grant no. 113964) lead by Professor Pascal Sciarini. The financial support of the SNSF is gratefully acknowledged. I would like to thank Pascal Sciarini, Simon Hug, Hanspeter Kriesi, Kenneth Benoit, Fabio Wasserfallen, Lisa Mu ller, Manuel Fischer, and the two anonymous reviewers for their valuable comments and suggestions. I am also grateful for the comments of Stefanie Bailer, Ursula H afliger, Susan Moffitt and various workshop participants at the University of Geneva. 2013 Swiss Political Science Association

Participation and Satisfaction with the Policy Outcome 61 argument goes as follows: The possibility to participate in decision-making processes gives individuals the feeling of being valued and that their opinion is heard and taken into account. This enhances an individual s procedural utility and leads to a more positive evaluation of the policy outcome (Frey and Stutzer 2005; Lind and Tyler 1988). Moreover, it has been shown that direct participation or the opportunity to participate can have a positive influence on satisfaction even if the actors do not fully approve of the outcome or if their economic benefit is low (Freeman and Langbein 2000; Olken 2010). Interest group research, on the other hand, has been largely concerned with power and influence. In these accounts, participation gives an actor the opportunity to influence the outcome. The more an actor is able to influence the outcome, the happier she is with the outcome. Thereby, the actual policy that is, the benefit of the new policy is important. However, access to decision-makers is only a prerequisite; whether an organisation is able to influence the policy outcome depends on group-specific characteristics such as resources and the type of interest it represents (D ur and De Bievre 2007b, Maloney et al. 1994). Recent studies also point to the importance of the decision-making context. Some issuespecific factors, such as the salience and the level of conflict of an issue, can be favourable to lobbying activities or make it more difficult for groups to achieve their policy goals (Mahoney 2007; Kl uver 2011). This study combines the two research traditions by studying two aspects of satisfaction with the policy outcome: first, the perceived level of consensus and, second, the perceived individual benefit for the group that is, how well the organisation s policy position is represented in the outcome. Our main argument is that both aspects of outcome satisfaction are positively related to the level of participation in formal stages of the policy process, but that the strength of the relationship is dependent on group- and context-specific characteristics. Our results show that public interest groups report higher satisfaction with the policy outcome, the more they participate. Further, we find an important effect of conflict: The higher the level of conflict of an issue, the less likely it is that the level of participation and outcome satisfaction are positively related. The study is based on a small but broad sample of policy domains. We analyse survey data collected through interviews with organisations involved in seven pivotal policymaking processes in Switzerland between 2001 and 2006. Swiss democracy is based on powersharing and is an ideal case of consensus democracy, according to Lijphart s (1999) famous classification. Legislative projects are prepared in a long, intense pre-parliamentary phase, during which actors try to find a consensus before the parliament even becomes involved (Kriesi 1980; Sciarini 2007). These preparatory stages offer various participation possibilities for all kinds of interest groups, and public and private organisations. Thus, Switzerland is an ideal case to study interest group participation in policymaking. In what follows, we discuss the different literatures concerned with benefits of participation in political decision-making and develop our hypotheses. Later, we present the Swiss policy process, our data and the operationalisation. The succeeding section summarises and discusses our findings. The final section concludes and suggests prospects for further research. Participation and Satisfaction with Political Outcomes Originally developed in the context of legal procedures, procedural justice theory claims that the opportunity to defend their interests ( voice ) is valued by disputants, even if they disagree with the outcome (Lind and Tyler 1988; Lind et al. 1990; Thibaut and Walker 1975). The psychological processes that produce the voice effect have essentially

62 Denise Traber been explained in two ways: Early theories claimed that an individual who is given the opportunity to express his or her views will have a sense of control over the outcome, which leads to higher procedural fairness judgements (Lind et al. 1990; Thibaut and Walker 1975). Later on, researchers increasingly advocated the intrinsic value of voice; that is, the personal utility individuals gain from participating in decision-making (Lind and Tyler 1988). A number of studies on participation in political processes have built on these psychological explanations of benefits of participation. Survey-based and experimental research has demonstrated the beneficial effects of political participation. Direct participation opportunities can enhance political efficacy (Bowler and Donovan 2002), improve the political knowledge (Benz and Stutzer 2004), make people happier with the way democracy works (Anderson and Guillory 1997), and as claimed by Frey and Stutzer (2005) improve their subjective well-being in general. Moreover, it was established that people care about procedure as much as they do about policies (Hibbing and Theiss-Morse 2001). Procedures that offer opportunities for personal involvement lead to higher satisfaction with procedure and outcome than representation, even in situations where participants do not fully agree with the outcome or if the outcome remains unchanged (Olken 2010; Esaiasson et al. 2012). Further, in the tradition of procedural justice theory, the American literature on bureaucratic policymaking has been concerned with participation opportunities of non-bureaucratic actors and how participation relates to the acceptance of policy decisions. In bureaucratic policymaking, participation is a means to legitimise policy decisions of public officials who are not elected and, therefore, not directly accountable to the public (Balla 2000; Golden 1998; Kerwin 2003). A number of studies show that participants appear to be more satisfied with the outcome in those procedures that offer better participation opportunities. Thus, satisfaction with the substance of the outcome as well as with the overall process is greater in negotiated rule makings, which are more open than conventional rule making procedures. Again, it appears that satisfaction does not fully depend on the net benefit of the outcome (Freeman and Langbein 2000; Langbein and Kerwin 2000). Despite these positive effects of participation, it is, of course, an established result that some participants tend to have more influence in policymaking than others. Indeed, much research on bureaucratic policymaking in the United States has been devoted to the question of who participates? and the responsiveness of bureaucracies to the requests of different participants (Balla 2000; Kerwin 2003). Several studies point to the advantage of business interests (Golden 1998; Furlong 1999). According to Yackee and Yackee (2006: 131), this bias can be explained by the quality of information provided by business interests: Public agencies are more likely to respond to the high quality information and scientifically sound technical data provided by the experts of businesses and industries. Contrary to the psychological explanations of procedural justice theory, interest group research (often implicitly) tends to relate outcome satisfaction to power and influence. Participation (or access) gives organisations the opportunity to influence policy outcomes and participants are happier the more they were able to draw the outcome towards their preferred position. Participation (or access) is generally seen as an important prerequisite for influence, as a necessary but not sufficient condition for satisfaction with the policy outcome (Maloney et al. 1994). Whether an organisation can influence policy outcomes depends to a large degree on its financial resources and whether it is able to provide the necessary policy expertise (D ur and De Bievre 2007b; D ur 2008a; Mahoney 2007).

Participation and Satisfaction with the Policy Outcome 63 Our study integrates the two research traditions. Focusing on participation in formal stages of the policymaking process, 1 we argue that participation should positively influence participants satisfaction with the outcome: First, because participation should give an actor some sense of control over the outcome and enhance the feeling that his or her view is being valued. Second, participation gives an actor the opportunity to influence the substance of the policy outcome. Influence should lead to a greater net benefit for the organisation, thus enhancing outcome satisfaction. To account for these two different mechanisms, we will study two different aspects of outcome satisfaction: perceived consensus and perceived preference attainment. Perceived consensus should be high if an actor thinks that the outcome is balanced and includes the interests of the various groups involved in the process. This points to a collective benefit of the decision. Perceived preference attainment refers to the individual benefit. Our first hypothesis postulates a general association between participation and outcome satisfaction. H1: Organisational representatives report higher satisfaction with the policy outcome the more their organisation had participated in the policymaking process. Second, we argue that the association will be stronger for some types of interest groups than for others. It is often discussed in the interest group literature that diffuse and specific interests differ not only in their prospective lobbying success but also in their participation opportunities and, therefore, also in the strategies they use to lobby decisionmakers. Diffuse interests associations such as environmental organisations, women s rights groups, or consumer organisations often lack a clear-cut constituency. They represent broad segments of society or defend a public cause. Specific interests, on the other hand, defend the concentrated professional, economic, or social interests of their members (Beyers 2004). These organisations often have well-endowed members who are willing to provide the necessary funding or expertise in view of the concentrated costs and benefits of the issue at stake. Organisations defending diffuse interests, on the other hand, find it more difficult to mobilise resources, as they are likely to face collective action problems because issues of public interest groups often do not have these concentrated costs and benefits for a specific constituency (D ur and De Bievre 2007a; Olson 1965; Schattschneider 1960; Schneider and Baltz 2003; Yackee and Yackee 2006). 2 In the logic of an exchange model in which the government offers influence in exchange of resources such as expert knowledge, technical advice, or implementation guarantees (Maloney et al. 1994; Kervin 2003), it is generally expected that inside lobbying strategies are reserved for actors who can provide these necessary resources (Beyers 2004). However, the distinction between insiders and outsiders is probably not as clear-cut as suggested by the model. Several studies point to the fact that interest groups combine different strategies and that most groups have at least a peripheral insider status (Binderkrantz 2005; Maloney et al. 1994). Beyers (2004) argues that policymakers are aware of the strategic information provided by specific interests and seek for more preference-neutral information by providing access to diverse organisations, including those that defend diffuse or public interests. 3 1 This includes consultation by and negotiation with decision-makers (Maloney et al. 1994) but also lobbying activities of groups, depending on the specific stage of the policymaking process. We will discuss the different stages of the Swiss policymaking process in detail below. 2 However, Pollack (1997) finds that diffuse interests can, in fact, be quite successful. 3 We use diffuse and public interests as interchangeable terms in this study.

64 Denise Traber Still, while specific interest groups especially those representing large industries will be consulted on different occasions, diffuse interests gain access only for specific issues where their expertise is needed or they can help improve the accountability of a measure. Thus, we argue that (formal) participation is more important for diffuse interests than for specific interests. First of all, specific interests often have other channels to gain direct access to decision-makers. Members of business interest groups and decision-makers are often part of the same networks; they might have met during their studies or earlier in their careers. These informal networks are especially important in a small country such as Switzerland (Kriesi 1980). Information exchanges within these networks can be more beneficial for an organisation s interest than formal participation in the policy process. Second, as organisations representing diffuse or public interests are less likely to gain access, they are invited only when their expertise is indispensable. In such situations, their participation should be more effective, which, in turn, should enhance their satisfaction with policy outcomes. We thus expect a stronger association between participation and reported preference attainment among groups defending diffuse interests than among those representing specific interests. We do not, however, expect such difference for the perceived consensus of a decision. H2: The positive association between the level of participation and reported preference attainment is stronger for diffuse interests than for specific interests. Apart from the type of interest, it is an established fact that powerful groups enjoy preferential access to the policymaking system. The concept of power has proved to be somewhat elusive and not easy to measure (Dahl 1957; Bachrach and Baratz 1962; Barry 1980). This study uses a slightly different concept, namely that of reputation or attributed power. Organisations with a high reputation are known for their influence in their respective policy fields and, therefore, are of great use for policymakers: First, because they are likely to provide the right kind of resources and, second, because they have means to secure support for a new policy among their members. As described below, reputation is especially important in Switzerland. Decision-makers have to consult powerful groups so as not to risk a referendum defeat of a new policy. However, we would again expect the association between formal participation and preference attainment to be stronger for groups with low reputation for the same reasons discussed above. Low-reputation groups take part less frequently in formal stages of the policymaking process most likely on issues that are most salient to their interests and, therefore, are likely to put much effort into their lobbying activities and provide useful information to decision-makers. Whereas groups with high reputation can use other (informal) channels, low-reputation groups rely much more on formal participation when they are granted access. H3: The positive association between the level of participation and reported preference attainment is stronger for groups with low reputation than for groups with high reputation. Besides type of interest and reputation, we know from earlier research that the extremity of an actor s preferences can considerably decrease his or her chances of success. Although empirical evidence is somewhat contradictory (see Bailer 2004), we assume that it is in general more difficult for an organisation to achieve its policy goals with a policy position too far away from those of other actors or from the status quo (Mahoney 2007). Similarly, it will be more difficult in a situation where an organisation favours the status quo, whereas the general consensus is on reform. An organisation that has conflicting interests will, in

Participation and Satisfaction with the Policy Outcome 65 general, face more constrains because it is less likely to find coalition partners. The ability to form coalitions is an important asset in lobbying. Being part of a large issue-specific lobbying coalition greatly enhances an organisation s chances of obtaining its preferred outcome, as large lobbying coalitions are more likely to succeed in lobbying (Furlong 1997; Baumgartner et al. 2009). For these reasons, we assume that if an actor s policy goals are in conflict with the preferences of other actors (perceived conflict), satisfaction with the outcome (perceived consensus and preference attainment) will be low, even with a high involvement in the decision-making process. H4: The positive association between the level of participation and satisfaction with the policy outcome is weaker the higher the perceived conflict. While interest group research has largely focused on group-specific aspects of influence, it has somewhat neglected the context within which the various lobbying, consultation, and negotiation activities take place. A few studies have recently shown that in addition to group-specific characteristics, such as resource endowment, the issue context matters for policy success (Mahoney 2007; Michalowitz 2007; Kl uver 2011). Success is much more difficult to achieve when the level of conflict is high; therefore, outcome satisfaction should generally be lower in these policy processes. While we do not have data for a stringent test of a context-level hypothesis, we will use a simple measure for the average level of conflict as a control variable (see Data and Measurement). Organisational Participation in the Swiss Policymaking Process The legislative process in Switzerland often begins in parliament with a parliamentary initiative, a motion, or a postulate. 4 Otherwise, the process can be initiated by a public vote on a popular initiative or referendum; in most cases, however, the government decides to pursue a new policy project or law reform (Sciarini 2007). In any case, the federal offices are in charge of designing new legislation or law reforms. In most instances, the office assigns the drafting of a new or revised legislation to one or several expert groups, consisting of interest groups and independent experts in the policy field. 5 Sometimes, so-called working groups are appointed in addition or instead of expert groups. By definition, working groups consist exclusively of members of the federal administration. In practice, however, non-bureaucratic actors often take part in working groups. Expert and working groups often organise hearings and invite organisations representatives to provide insight and practical advice on specific policies. After preparation of a first draft within expert or working groups, the legislative project is submitted to a consultation procedure (Vernehmlassung). During a certain period, the public is asked to comment on the proposed legislation. Major parties, interest groups, and experts are officially contacted by the administration, but any organisation and citizen is allowed to formulate his or her position on the subject. The consultation is usually closed after about three months. Thereafter, the office in charge prepares the draft bill, which is passed on to the parliament. 4 The parliamentary initiative is a draft of a new law authored by a member of parliament or a party group. By motion (binding) and postulate (non-binding) the parliament demands a new law or specific revisions from the government. 5 For a more detailed description of the legislative process, see Sciarini 2007.

66 Denise Traber A couple of weeks before the parliamentary session begins, the legislative committee in charge 6 meets and prepares its own draft of the legislative project, which can differ from the government proposal. Legislative committees organise hearings and consult experts in the policy field. They invite interest groups and seek advice from various private organisations. It is generally assumed that specialists from organisations or business have considerable possibilities to influence the decision-making process at this stage, mainly by providing expert knowledge, often lacking among the non-professional MPs in Switzerland. The parliament meets at least four times a year for three consecutive weeks. During floor debates, formal access to the political process is restricted to members of parliament. However, several areas of the parliamentary building are accessible to journalists and lobbyists. Moreover, interest groups often invite members of parliament to special events, such as dinners or presentations. Contrary to other democracies, the legislative process in Switzerland does not end with the parliamentary decision. A large proportion of the total number of legislative acts is subject to a popular referendum, which can be mandatory or facultative. 7 A facultative referendum takes place if 50 000 signatures are collected against the new legislation within 100 days after the law s publication in the federal law gazette. By and large, the legislative process in Switzerland offers formal access to interest groups at various stages. The pre-parliamentary policymaking process in Switzerland is somewhat comparable to administrative rule making in the United States. The bureaucracy gives access to various private organisations to maximise legitimacy of a new rule. However, unlike policymaking in other democracies, broad acceptance of political decisions is a real necessity in Swiss politics. Direct democracy is one of the most important elements of the political institutions and political culture (Kriesi 1999). Decision-making outcomes are difficult to predict. So as not to risk a referendum defeat, the government is forced to include at least those groups that have the resources to call for a referendum (i.e. collect the 50,000 signatures) (Lehner 1984; Neidhart 1970; Papadopoulos 2001). The general political culture in Switzerland puts high value on participation of the public (individuals and groups) in political processes; therefore, it is rare for groups to focus exclusively on outside lobbying strategies. Whenever they are granted access to policymaking in areas important to their constituency, interest groups try to become involved as early in the process as possible; blocking legislation at the referendum stages is only a last resort and not a very efficient strategy. So, interest groups will participate in expert committees when invited and present their views in front of parliamentary committees because these participation possibilities are much more effective than outside lobbying. The bureaucracy, on the other hand, is forced to hear and include different views on the subject, so as not to risk a public referendum (Lehner 1984). We can therefore assume that the decision to participate (when granted access) is not influenced as much by the expected outcome as it is the case in other, more exclusive, political systems (for research on the US case, see Balla 2000). 6 In the majority of cases legislative projects are assigned first to the Upper House. 7 The referendum is mandatory for constitutional amendments, accession to international organisations, and urgent bills; it is facultative for general bills and for treaties under international law (Bundesverfassung der Schweizerischen Eidgenossenschaft vom 18. April 1999, Art. 140).

Participation and Satisfaction with the Policy Outcome 67 Data and Measurement The data were gathered as a part of a larger research project on legislative decision-making in Switzerland (Sciarini and Serd ult 2006). To select our cases, we first conducted a survey among approximately 80 experts of Swiss politics. The experts received a list of all laws passed by the Swiss parliament between 2001 and 2006, on which a referendum vote was allowed (regular bills and international treaties) or required (changes of the constitution). 8 This list included about 350 laws. In about 11% of the cases a popular referendum had actually taken place. The experts were asked to assess the importance of each law on a five-point scale. Based on these expert judgements, we selected the ten most important laws. For this present study, we focus on policy processes dealing exclusively with national legislation, which are relatively homogeneous in terms of rules of procedure. After excluding international legislation, we cover the seven most important policymaking processes concerning national policy in various areas, such as social policy, migration, spatial planning, and energy policy (for an overview, see Table 1 in Appendix). Our data were gathered during interviews with representatives of organisations involved in these seven policy processes. 9 The organisations were selected on the basis of detailed chronologies of the decision-making process. More specifically, we listed the most important formal stages and prepared a list of all organisations that had participated in at least one stage of the process besides the consultation procedure. Thus, our sample includes all organisations that participated in at least one stage of the most important national policymaking processes between 2001 and 2006. While some of the organisations were present in several processes, others participated in only one. Our questionnaire included a survey and open questions. The variables described below were constructed on the basis of the survey; we will later use some qualitative information also gathered during the interviews to illustrate our findings. Satisfaction With the Policy Outcome We use two different measures of satisfaction with the policy outcome. The first measures perceived consensus; more precisely, whether the different interests were equally represented in the outcome or whether certain interests prevailed at the expense of others. This measure grasps an evaluation of the collective benefit of the policy outcome. In a consensus democracy, institutions are designed to accommodate the different interests of a heterogeneous society (Lijphart 1999). The collective benefit should be higher, the better the various interests are represented in policy outcomes. During the interviews, the representatives were asked to indicate on a five-point scale whether they thought that the outcome had been determined by narrow interests (1) or was decided by consensus (5). The measure was rescaled for comparability with the second dependent variable. The distribution of the main variables is displayed in Figure 1 in the Appendix. 10 8 Urgent laws with validity of more than one year are subject to a mandatory referendum if they lack a constitutional basis, otherwise the referendum is facultative. 9 The necessary information for the variables described in the following was obtained during semi-structured interviews with representatives of these organisations between February and August 2008. The interviews were conducted in French and German. 10 We allowed the interviewees to give intermediate answers (e.g. 2.5). Both measures of satisfaction with the policy outcome therefore contain more than four categories, as shown in Figure 1 in the Appendix.

68 Denise Traber Second, we use a more direct measure of the perceived individual benefit for the representative s organisation. This variable measures to what degree in the view of the representative the organisation s policy goals were attained. It can also be seen as a measure for perceived influence; however, to avoid the discussion of influence (actors could be just lucky and obtain their preferred outcome without much effort on their part; see Barry 1980), we call this measure perceived preference attainment. 11 Our questionnaire contained a list of the most contested issues in each policymaking process, created by means of content analysis of parliamentary debates and various documents provided by the federal offices (see Table 1 in Appendix). The representatives were asked to rank the policy outcome for each issue on a four-point scale. More specifically, they had to evaluate for each issue whether, in their opinion, the organisation s interests had been taken into account; (answer categories: not at all (1), rather not (2), a little (3), fully (4)). 12 At the same time, the representatives had to indicate the importance of each issue. 13 The measure of perceived preference attainment takes only salient issues into account. 14 More precisely, we discarded all issues that were regarded as unimportant and calculated the median preference attainment for the important issues per organisation and process. The two dependent variables are highly correlated (r = 0.7), even though they were measured very differently. While perceived consensus was assessed directly by a single research question in the questionnaire, the second variable preference attainment was calculated on the basis of several questions. With these two measures we should be able to account for different aspects of outcome satisfaction: A more general evaluation of the decision in terms of balanced interest representation and a more specific, individual evaluation of the benefit for the organisation. Participation Our main independent variable is participation. Our questionnaire allowed us to assess specific participation in certain stages of the policymaking process as well as the overall degree of participation. 15 Each decision-making process was divided into several stages. The representatives had to indicate for each stage (approximately 15 for each law) whether their organisations had been involved in this specific stage or not. We aggregated the substages into the following four categories: decision-making within the office(s) in charge of the policy, expert and working groups, parliamentary committees, and activities during parliamentary sessions. 16 Further, the variable degree of participation counts all stages 11 Note that many studies use the same term to describe an objective measure; that is, they measure the distance between the organisation s preferences and the final policy outcome (Schneider and Balz 2003; D ur 2008b). Contrary to these studies, we use a subjective measure of preference attainment. 12 The questionnaire included a short description of each issue, as well as a short summary of the outcome (corresponding section of the law) for each issue. 13 Answer categories: unimportant (1), rather not important (2), somewhat important (3), very important (4). 14 For a discussion of the role of salience in decision-making, see Thomson et al. 2003. 15 We can, however, only indicate whether an organisation had participated in a specific phase or not. Unfortunately, we do not have information on the type or intensity of participation within this specific stage. As discussed in the last section, the type of participation follows the rules within the different stages of the formal decisionmaking process, and the type and degree of participation within these stages should therefore not vary too much. 16 These four categories capture the most important stages of the legislative process in Switzerland, except for the referendum (Kriesi 1980). Because satisfaction with the outcome was defined as satisfaction with the parliamentary decision, it does not make much sense to include referendum participation. The consultation procedure was not taken into account because all groups participated in this stage.

Participation and Satisfaction with the Policy Outcome 69 (including sub-stages of the aforementioned categories) in which an organisation participated, divided by the total stages in the process. This continuous variable takes values between 0 (no participation) and 1 (participation in all stages of the process). 17 Figure 1 in the Appendix shows the distribution of this variable in the sample. It shows that even if the sample contains no groups participating at all, certain groups have a very low level of overall participation and the distribution is approximately normal. Type of Interest The organisations were divided into four categories (dummy variables), according to the type of interest they represent. We distinguish between representatives of local (cantonal) government, business interest groups, unions, and public interest groups. Swiss cantons are quite autonomous in certain policy areas, such as social policy and education. Thus, to defend regional interests, the associations of cantonal executives often appear as lobbying organisations. The Association of Social Ministers (SODK), for example, plays an important role in social policy. This category also includes the associations of Swiss municipalities and Swiss cities. The category public interest groups includes several organisations that defend public or diffuse interests, such as environmental organisations, women s groups, refugee s rights organisations and consumer protection groups. We will use public interest groups as a reference category in all statistical models. The category business interest groups subsumes general business interest, as well as groups defending specific interests such as the Swiss Touring Club (a motor club), and representatives of the electricity industry. Like public interest and diffuse interest, we use business interest and specific interest as synonyms in this study. Reputation The reputation or attributed power of an organisation was measured as follows: We presented a list of approximately 150 organisations to the representatives and asked them to judge the general importance of each organisation for Swiss politics between 2001 and 2006 (unimportant = 0, important = 1). We pointed out to our interviewees that they should judge the importance irrespective of a policy area or a specific policymaking process. The variable calculates the fraction of all interviewees who named the organisation as powerful in a continuum between 0 and 1; for example, a value of 0.6 signifies that the group had been tagged as powerful by 60% of all respondents. Because the distribution of this measure proved to be essentially bimodal (see Figure 1 in Appendix), it was converted into a dummy variable, which takes the value 0 if the reputation is below or equal 0.5 and the value 1 if the reputation is above 0.5. Perceived Conflict During the interviews, we presented another, much shorter list that included only the organisations that participated in the respective process. The representatives had to indicate for each organisation on the list, whether there was agreement, disagreement, or neither between the two organisations during the decision-making process. The variable perceived conflict counts the number of organisations with which the representatives indi- 17 The referendum campaign was again excluded.

70 Denise Traber cated conflict, divided by the total number of organisations in each process. This variable measures perceived conflict of an interest group with all groups participating in a specific policymaking process, including parties. The variable has a minimum of 0 (no conflict) and a maximum of 0.55 (conflict with 55% of organisations). The distribution is again displayed in Figure 1 in the Appendix. Level of Conflict in Decision-Making Processes This variable is a measure of the average perceived conflict in each process. Unlike the variable perceived conflict, which was measured on the level of organisations, the level of conflict accounts for an important characteristic of the decision-making context; namely, the general agreement/disagreement of all non-state actors (including parties). More specifically, this variable calculates the average perceived conflict of all participating actors including parties, and is more comprehensive than a simple average of the actors in the sample would be. 18 It should thus accurately represent an aspect of the decision-making context. 19 Table 2 in the Appendix gives an overview of the level of conflict in each process, as well as the average degree of participation. It appears that those processes with the lowest level of conflict (Constitutional Article on Education and Infrastructure Fund) were also the most inclusive. Limitations of the Dataset Note that the study is based on a sample of actors that were selected according to various criteria. The dataset contains only actors that had actually participated in at least one stage of the policymaking processes; therefore the study compares actors with different levels of participation. As shown in Figure 1 in the Appendix, however, the variation in the degree of participation is reasonably high. Moreover, we argue that the self-selection mechanisms are not as important in Switzerland as they are in other democracies. Arguably, actors should only decide to participate if the costs of participation are less than the expected benefit, or put differently, if they can expect the outcome to move closer to their preferred outcome because of their participation. As discussed above, in Switzerland the administration invites interest groups to participate in pre-parliamentary policymaking, in order to find a consensual solution and prevent a popular vote. Interest groups thus always have an incentive to participate when invited and even more so when very important laws are at stake as it is the case in the policymaking processes included in our sample. This being said, our selection criteria as well as the fact that the study includes variables that are based on retrospective perceptions may nevertheless lead to an endogeneity problem, which we are aware of. We nevertheless think that our results are interesting for those actors that actually participate in political decision-making. Moreover, our results reveal 18 Number of cases included in the variable level of conflict: Pension reform: N = 16. Nuclear energy act: N = 19. Financial equalisation: N = 17. Foreign nationals act: N = 14. Education: N = 16. Telecom: N = 18. Infrastructure Fund: N = 17. The sample we use in our analysis is much smaller for two reasons: We excluded parties and some organisations dropped out due to missing values. 19 Another measure for the level of conflict would be conflict in parliamentary votes on the seven laws. We refrained from using this measure for two reasons: first, it would only take into account conflict among parties, second, we expect conflict to be lower at the final voting stages of a process, and since our study is concerned with the whole policymaking process, this would not be an accurate measure of the level of conflict.

Participation and Satisfaction with the Policy Outcome 71 important differences in reported outcome satisfaction between groups with high and those with low participation, as well as between different types of interest groups. Results Who Participates and to What Extent? Satisfaction with policy outcome might be unequally distributed because some groups participate more than others. The literature generally assumes that business interests gain preferential access to the political system (Golden 1998; Yackee and Yackee 2006). To assess whether this is the case in Switzerland, we calculated the average participation in different stages of the policy process and compare those figures with the average outcome satisfaction in Table 3. We distinguish four different stages that offer different participation opportunities: participation in decision-making within federal offices (intra-administrative stages), participation in expert committees and working groups, admission to parliamentary committees, and access to decision-makers during parliamentary sessions. The first four columns of Table 3 show the percentage of organisations (within group type) that had access to the respective stage of the policy process. 20 It is not very surprising that local government officials have the highest participation in all phases except the parliamentary stages. As discussed above, cantonal ministers are organised in policy-specific associations that often appear as lobbying organisations. However, as representatives of sub-national entities, they also enjoy preferential access to intra-administrative stages. Comparing the non-bureaucratic actors, we find considerable differences between business interests and the other groups, which is in line with earlier results (Linder 2005, Kriesi and Trechsel 2008). Business interest groups not only show the highest average degree of participation, but they also enjoy preferential access to expert/working groups and to parliamentary committees. Another striking result in Table 3 is that almost 30% of public interest groups (diffuse interests) participated in intra-administrative stages (compared to 20% business and only 10% unions), and more than 60% of these organisations were active during the parliamentary sessions. This supports our earlier assumption that diffuse interests can be important for the administration in certain cases on these specific occasions they might enjoy preferential access to the administration. Their high level of participation during parliamentary sessions confirms yet another expectation: These groups are much more active in lobbying parliamentarians than other groups possibly because they focus on specific legislation, participating less often but putting much more effort into their lobbying activities on these occasions. The summary figures for groups with low and high reputation (bottom two rows in Table 3) seem to contradict previous research. We would expect groups with high reputation to participate more than those with low reputation because of their power to call for a referendum (Lehner 1984; Neidhart 1970). However, according to the summary statistics in Table 3, low-reputation groups are more intensely involved in Swiss policymaking. These figures again suggest that groups with low reputation participate less in general but are highly involved in cases where they can gain access to the administration. 20 Note that our measure counts only whether an organisation had access at least once during a specific process. We can not assess the intensity of participation within a specific phase, nor can we distinguish between active members of expert groups and those organisation that were invited only once to a hearing.

72 Denise Traber Table 3: Participation and Satisfaction with Outcome by Group Type Participation Satisfaction with Outcome Federal Offices (%) Expert Committee and Working Group (%) Parliamentary Committee (%) Parliamentary Session (%) Participation Degree (Mean: SD in Parentheses) Perceived Consensus (Mean (SD)) Preference Attainment (Mean (SD)) N Local Government 0.58 1.00 0.92 0.42 0.63 (0.26) 3.00 (0.87) 3.40 (0.76) 12/9 Business Interest 0.21 0.79 0.93 0.43 0.53 (0.26) 3.00 (0.80) 3.20 (0.61) 14/8 Union 0.10 0.50 0.60 0.30 0.32 (0.25) 2.50 (1.20) 2.40 (0.94) 10/3 Public Interest 0.29 0.57 0.79 0.64 0.46 (0.17) 2.50 (1.01) 2.50 (1.07) 14/13 Reputation > 0.5 0.17 0.61 0.72 0.39 0.42 (0.28) 2.90 (1.00) 2.90 (0.82) 18/5 Reputation < 0.5 0.38 0.78 0.88 0.50 0.53 (0.23) 2.70 (0.97) 2.90 (1.03) 32/28 Note: The first four columns show the percentage of organisations (within group type) that participated in the respective stages of the process. Columns 5-7 show the average participation degree and average satisfaction within group type (standard deviation in parentheses). The last column lists the number of organisations in the dataset. Organisations that participated in several processes are counted each time as individual observation. We list the number of observations as well as the effective number of organisations. For example, there are 8 different business interest groups in the sample, but 14 observations in total, because three of these organisations were involved in several processes.

Participation and Satisfaction with the Policy Outcome 73 High-reputation groups, on the other hand, have much broader presence and are granted access on different occasions due to their expected power but also due to their broader expertise. 21 Considering satisfaction with the outcome, groups do not seem to differ much, at least when looking at the averages. However, unions and public interest groups have much larger standard deviations, especially with regard to perceived preference attainment, pointing to an advantage of business interests and local government officials. Interestingly, business interests and local government report higher preference attainment than perceived consensus. Some representatives quite openly admitted that in some cases consensus was low but the outcome was at the same time very favourable to their organisation. 22 Type of Interest and Participation Our hypotheses postulate that satisfaction with the policy outcome should be higher the more a group participates in the policymaking process. Further, we posit that the relationship between participation and outcome satisfaction should by stronger for some types of groups than for others. To test our hypotheses, we estimated different models with ordinary least squares (OLS). The regression coefficients and corresponding standard errors are displayed in Table 4. 23 Models 1 4 compare different types of interest groups. 24 As discussed, we distinguish between local (cantonal) government, business interest groups, unions, and public interest groups. The reference group is public interest groups in all models. Because of the interaction terms, the coefficient for degree of participation is the value of the reference group (i.e. public interest groups). 25 It appears that diffuse interests are indeed more satisfied with the outcome the more they participate, which supports H2. Models 1 and 2 predict a 3-point higher satisfaction for full participation (100% of all stages), compared to no participation for public interest 21 See also the last column in Table 3, which shows the number of observations and the number of different organisations within group type. Several of the business interest groups in our sample were present in several policy processes, while the number of different public interest groups almost equals the number of observations. Similarly, most low-reputation groups participated in only one process, while all high-reputation groups were present in several processes. 22 For example the organisation of justice and police (KKJPD) reported a very high level of preference attainment (4) but a low level of consensus (1.75 rescaled) for the Foreign Nationals Act. Similarly, the Swiss Organisations of Employers (SAV) admitted that consensus was very low in the 11 th Pension Reform (1.75 rescaled), even though their preferences had mostly been taken into account (3). 23 The models were run with cluster-robust standard errors. Because of the low number of cases, and especially in some categories of the dependent variables, we do not use ordered logit/probit, even though we are aware that this would be the correct statistical procedure. However, we recoded the variables into four categories each and ran ordered logit for a test, finding almost the same results as with OLS, leading to the same conclusions. Results are available upon request. 24 The models are estimated separately for each dependent variable. We also tested joint estimation of the model pairs; the estimation by SUR models leads to almost identical results. 25 We do not include perceived consensus in the regression on preference attainment for several reasons. Arguably, we would assume that perceptions of the decision-making procedure would affect an actor s perceived preference attainment. However, our measures of perceived consensus and preference attainment measure different aspects of the outcome and while both are certainly related, it is difficult to argue which one causes the other. Moreover, including perceived consensus in the equation for preference attainment (or vice versa) would absorb most of the variation in our small sample and we therefore refrain from doing so.