Towards a systematic account of constitutional success and failure Constitutional change in Switzerland

Towards a systematic account of constitutional success and failure Constitutional change in Switzerland Dietmar Braun, Raffael Himmelsbach, Björn Uhlmann Institut d Etudes Politiques et Internationales, Université de Lausanne Paper presented at the ECPR Joint Sessions 2008 in Rennes. Workshop 20: The Politics of Constitutional Change 1. Introduction Our objective in this paper is to identify relevant variables for explaining success and failure in amending constitutions in the context of federalism. We will therefore, first, not discuss constitution making from the scratch, i.e. the construction of a new constitution, for example after a transition from an authoritarian to a democratic regime. The context for amending constitutions is different: it is usually based on a stabilised democracy with established and legitimised rules and procedures. And in distinction to a new constitution, amending constitutions means to overcome an established status quo, i.e. existing rules or procedures in favour of a new rule or procedure. In a context of transition, such rules or procedures must yet be defined and need not be delimited against the existing order. We will, second, treat the issue in the context of federalism and choose only one federal country, Switzerland, as an exploratory case which introduces regional identity questions and the territorial distribution of powers as relevant additional variables into the equation of constitution amendment. A commonly accepted analytical model of constitution making is not yet at hand despite of numerous efforts to describe constitutional change in the context of individual countries. We will not be able to deliver a completely designed model either in this paper. Our more modest intention is to make first exploratory steps towards a causal model of constitutional success and failure by identifying relevant and irrelevant causal dimensions and variables that interact in specific ways. This way of seeing things is

2 inspired by Ragin s multiconjunctural causation methodology ((Ragin, 1987)) and Goertz and Mahoney s two level theory (Goertz and Mahoney, 2005). Ragin s method was developed to take the interplay of causal variables into account that only in certain combinations result in a certain outcome. Goertz and Mahoney stress the possibility of several levels of causation within causal models and as Ragin the relevant distinction between necessary and sufficient causal conditions. The development of a more complex causal model of constitution making would be a first step towards more systematic research. We will proceed as follows: The next section serves to sketch our preliminary causal model for the explanation of successful constitutional change. The three case studies of constitutional amendments in Switzerland that follow are used as empirical material to confirm or disconfirm our expectations concerning the causal variables in the model. A discussion of empirical results follows which will lead to the final explanatory model. Conclusions wrap up the argument. 2. Outlining the causal dimensions of the explanatory model On the base of previous work (Braun, 2008) and an analysis of comparative literature on constitution making in federal countries (e.g. Banting and Simeon, 1985b) we think that the following analytical dimensions should be taken into account in order to understand successful amending of constitutions. (1) Our first assumption is that conflict intensity is a primary causal factor that determines opportunities and difficulties to successfully deal with amending constitutions. The more intense the conflict around the issue the more difficult it becomes to find an agreement and the more it needs intense agenda crafting to overcome the tensions between actors. We think that there are three kinds of possible problems that need to be dealt with: Efficiency problems may arise if existing regulations, structures or procedures that are fixed in the constitution cause political failure in one way or another. The existing causal model or constitutional theory (Vanberg and Buchanan 1989) comes under pressure and needs to be replaced by alternative models. In general, if an issue is treated as an efficiency problem, conflict intensity can be low because all actors have a common interest in solving political failure. As long as

3 there are no contrasting visions or alternatives that are put forward to solve efficiency problems, this kind of problem is quite manageable. (Re)distribution problems can have considerable salience. Such problems emerge if the issue in question demands a distribution or redistribution of competences or financial resources that break with the status quo and can cause winners and losers. Value problems can also have strong salience. These are problems that are built on cleavages in the party system or in the territorial system and which are intimately linked to normative questions. They can be of different quality: some issues, like equity, may cause a mobilisation on the base of the left right cleavage, which usually polarises the party system. Others may stir identity problems, resulting in serious conflicts between the federal state and member states. The rise of other value cleavages is also possible depending on the mobilisation capacity of actors and the issue at hand. We see these problems as an exhaustive list of the kind of problems that can arise. An issue may raise all three problems at the same time. Actors can use agenda crafting to transform redistribution or value problems into efficiency problems or make efficiency problems the predominant frame of discussion in order to overcome resistance. How indeed political entrepreneurs develop strategies to strengthen the efficiency dimension and to depoliticise value and distribution conflicts, is one of the most interesting and relevant questions for understanding the success and failure of amending constitutions. (2) Next to it we find the role of interests to be defended and the number of veto players. This is obviously an important analytical dimension that needs to be taken into account. It is important to see if there are dominant actors that can easily prevail in the process, which kind of winning coalitions can be constructed or how different interests of participating actors are. Both conflict intensity and veto players are on a primary level within a causal flow model that one could develop here. On a next, secondary or intervening level we find three other analytical dimensions that are of relevance for the ability of political systems to deal with conflicts around constitutional issues: institutions, ideas, and process.

4 (3) Institutions structure three things: They define the access to constitution making (who becomes a veto player), the obstacles to overcome in order to amend the constitution (decision rules to amend the constitution; constitutional rigidity or flexibility), and the procedures to be followed. The time it takes to adopt an amendment and the possibility to construct a consensus are structured significantly by the kind of institutions in a country. (4) Ideas can have an important function in building a consensus despite of high conflict intensity. If actors share a common vision of the problem and solutions at hand because they believe in the same causal story that is constructed to discuss the issue at hand, it becomes easier to solve underlying tensions in interests between actors. Garrett and Weingast point to ideas as focal points, notions that are used in consensus building in order to reduce uncertainty about future opportunism of actors or, in other words, about the respect of the contract in the future (Garrett and Weingast, 1993; see also Weingast, 1995). The important point is that discussions can be framed, i.e. be influenced in a certain way by actors in order to create such a shared vision or focal point. In this way, ideas are a means to be exploited by actors. If they play a role depends, of course, on a large number of factors. (5) We see the process and the time dimension as key variables in our causal model. This is a dimension, which is often neglected. Veto player theory for example is completely based on a static vision of political decision making. Changes in preferences are usually not considered possible. We believe, however, together with Hart (Hart, 2001) that constitution making is a process of conflict transformation. It is a forum for negotiation and continuing conversation in which actors exchange points of views and judge on alternatives. Arguing obliges actors in difference to bargaining to refer to the common good as the main outcome variable and to find support by way of the better argument instead of referring to selfish interests (Elster, 1991). It is during this process of arguing that preferences may be reformulated and adapted to the alternatives that emerge and change constantly. Participation in the process of constitution making allows, in addition, also changing social relationships between actors: the constant exchange of visions and the elaboration of workable alternatives may found relationships based on trust instead of distrust and may also keep actors hostage of the process: the more actors play the game and try to elaborate solutions the more they are considered as being responsible for the solution and the more actors are becoming

5 obliged to demonstrate loyalty to the collective solutions that are discussed and found in the discussion process. All this suggests that the process should be treated as an important explanatory dimension in constitution making. If one accepts this view, one can also understand Riker s emphasis on the heresthetics of constitution making (Riker, 1984). A process allows the manipulation of social choice situations. It introduces an element of artistry : during the process of constitutionmaking actors have the possibility to invent, revise and eliminate alternatives. This also means that agenda s can be designed or manipulated according to interests. Agenda crafting may be a useful tool in building a consensus. This is why Riker votes for a dynamic model of constitution making. There are a number of different process variables that seem of relevance to us. (a) The first process variable refers to different phases or arenas of constitution making that can be distinguished (Banting and Simeon, 1985a). We distinguish between the arena of idea generation, the arena of agreement reaching, and the arena of ratification. It makes sense to use this distinction because institutional procedures and participating actors can differ in each arena, as does the overarching function of the arena. Actors participating in such arenas have therefore different rationales that are adapted to the function of the arena. This can be used by agenda crafting, e.g. by limiting idea generation to one causal story in the mobilisation arena in order to reach a consensus in the decision making arena. Moreover, one can also distinguish between different kinds of discourses that are predominant in these arena s. According to Schmidt (Schmidt, 2000) two reform discourses prevail: one coordinating discourse that serves to build a consensus among the political elite and one communicative discourse that searches to find acceptance of political solutions among citizens. While the coordinating discourse will prevail within the decision making arena, the communicative discourse is the main discourse in the ratification arena. The type of discourse in the idea generation arena is less straightforward: one could of course assume that the coordinative discourse is the predominant feature in this arena but it is clear that we may find a still different kind of discourse that is based on the integration of (scientific) experts into the process and may be labelled a rationalising discourse. Experts are included in order to develop convincing causal stories but they do usually not share the logic of actors participating

6 in the coordinative discourse and that serves to find a consensus among the political elite (Sciarini, 2004). The logic of a rationalising discourse and a coordinative discourse may be in conflict within the idea generation arena. (b) Taking up this discussion we would content that the second process variable is the integration or non integration of experts. Experts will usually be found in the idea generation arena but may also have their role in the decision making arena and even in the ratification arena. The political elite may use experts to overcome gridlocks in the discussion and experts can rationalise discussions and introduce new arguments that help to create shared visions on problems. (c) The third process variable is the inclusion or non inclusion of actors within the process of constitution making. This is generally recognised as one of the most important variables for the successful management of amending constitutions. Exclusion of certain veto players in the construction of a consensus can create serious resistance during the decision making or ratification phase. Inclusion may cause difficulties in finding workable solutions but at the same time actors have voice and they are drawn into the process of constitution making and become a responsible part of constitution making. As already said, the inclusion of actors may vary according to the three arenas and it may be part of the heresthetics if the constitution allows so to use inclusion or exclusion in order to improve the chances for the acceptance of the constitutional amendment. (d) The fourth and last process variable would be agenda crafting itself, the manipulation of the process in order to achieve a favourable result. Agenda crafting is usually underestimated. Our previous work points to the importance of timing, sequential ordering, selection of actors, integration of experts etc. The separation of the discussion on general principles and distribution questions can be part of a success strategy of agenda crafting. There is no need to develop this variable further at this moment and we will not be able to work it out in detail in our case studies. It is, however clear, that artistry or agenda crafting can be used and is used to achieve conflict transformation (Hart 2001).

7 Figure 1 Causal dimensions in constitution making The figure resumes the analytical dimensions that we think are important for developing a causal model of constitutional change. While we conceive of conflict intensity and actor interests and actor constellations as first order dimensions, institutions, ideas, and process have the status of second order dimensions that intervene. We have not yet developed clear expectations and testable hypotheses respectively nor have we elaborated in a more detailed way the kind of causal variables that may play a role in each causal dimension. Our explanatory study serves exactly to identify such variables and to allow for more precise causal statements concerning the influence of the causal dimensions and their variables. How do we want to do that? 3. How to proceed The elaboration of all dimensions of the causal model would be too complex and spacedemanding for this paper. As we wish to accentuate the process dimension we have decided to keep constant two of the causal dimensions: the structural/institutional dimension and the veto player dimension. In order to keep the institutional constant, one should not compare different institutional contexts in countries but different constitutional projects in one country, if possible also at about the same time in order to avoid biased results because of different historical contexts. Of course, if one wants to draw some generalising conclusions, one should be aware of the specificities of the institutional context and in how far this might structure the process of constitution

8 building. This is why we will start with a short chapter on the specific institutional context of Switzerland and the way it structures constitution making. As we will deal with three cases that concern amendments of constitutional articles dealing with federal relationships, the kind of veto players also remain usually the same in all cases. This allows in a similar way to avoid dealing with expectations on this dimension. We have selected three case studies of constitutional reform in Switzerland in the period 1990 2006: the reform of competence assignment and fiscal equalisation between the federal government and member states (adopted by a referendum in 2004 after 15 years of constitution making); the adoption of new constitutional articles on education and research in 2006 (after nine years of constitution making); and the refusal of a new article on the promotion of culture by the federal government in 1994 (after 13 years of constitution making). All three issues touch upon federal relationships and can therefore generate similar problems and mobilise the same actors, which helps, as said, to reduce the interference of variations on the actor dimension as an explanatory dimension. But there is also variation in the outcome: we have two successful cases of constitutional amendment (equalisation and education) and one failure (culture). This should give us the occasion to highlight explanatory variables that were present or absent and to stress the favourable and unfavourable combination of causal factors. We will start with a brief account of the institutional context before we give a summary account of the cases. 4. Institutional context What are the institutional particularities of Switzerland in dealing with constitutional change? If one refers to the well known typology presented by Lijphart (Lijphart, 1999), Switzerland is the extreme case of a consensus democracy polar opposite to New Zealand and the United Kingdom as majoritarian cases. Both on the political dimension and the territorial dimension of Lijphart s typology the search for integration and consensusseeking prevails: this is for example demonstrated by the oversized government coalition of the four largest parties being in office since 1959, the multi party system and the dominance of the principle of proportionality in the distribution of power on the

9 one hand and by the type of administrative federalism (Hueglin and Fenna, 2006), which in many policy areas links the federal level to the cantonal level and vice versa on the other hand (see also Benz, 2002). There are several theories about the origins of the consensus democracy in Switzerland. One of the explanations and this is a particular feature of Switzerland is certainly the influence of direct democracy on the political game (Neidhart, 2002): as direct democracy establishes a system with many potential veto players, it becomes primordial for political actors to find an encompassing consensus during the political decision making process. Integration of potential vetoplayers during this process reduces the probability of contestation during the ratification process or increases the chances that the law proposal will be accepted by a majority in the population despite of contestation. Integration of veto players may be the outcome of direct democracy or be built on a long history of government cooperation already at the level of communes (Lehmbruch, 1993), the result is a political culture and political practice that is built on balancing forces. This is reflected in the formal procedures that are needed to revise the constitution: There are two chambers, the National Council and the Council of the States, the last one representing formally the people of the 26 cantons. The formal rights of both chambers are completely equal and as in the US Senate it needs approval of both chambers to adopt a law or a constitutional amendment. Administrative federal cooperation is often organised informally though, especially in the field of higher education, official coordinating bodies are emerging in the last years. Switzerland has extensive pre parliamentary procedures that are used especially in the case of constitutional amendments (Sciarini, 2004). In the idea generation phase, which is formally not prescribed, a pre consultation of stakeholders (interest groups; cantons) can take place to test the potential support, applicability and validity of the proposal that is presented by the federal administration. In the agreement reaching phase a formal and prescribed consultation takes place where stakeholders have the possibility to react by written comments to the proposal. Only after this stage the message of the

10 government, i.e. the official law proposal or constitutional article, is presented to parliament. The final stage is the acceptance in the referendum arena. Each constitutional amendment must be submitted to a referendum and needs the majority in the population as well as the majority among the cantons (14 out of 26). If one makes abstraction of the time consuming integration procedures, constitutional change is comparatively easy in Switzerland and it is often done: propositions may come from the government, the parliament or interest groups/citizens if a certain number of signatures can be presented. Given these institutional and cultural features, is Switzerland a case that can serve to generalise knowledge on constitutional change? If one considers Switzerland in Lijphart s terms as the exemplary case of a consensus democracy, it should be possible to reveal also in an exemplary way the main features of constitution making in this type of democracy. This means on the other hand that conclusions might not be applicable to majoritarian democracies. In addition, Switzerland has however the character of a referendum democracy. Referendums are taking place in other countries but there is no comparable case that builds its democratic system as much on direct democracy as Switzerland. This does not mean that Switzerland is an outsider case. There are two consequences of this particular feature: First, the pressure to build an elite consensus before the referendum takes place might be higher than in parliamentary democracies but the building of an elite consensus larger than normally is also present in these democracies as it often needs supermajorities to adopt constitutional amendments. Second, Switzerland has to pass two steps instead of one as in the parliamentary democracy: building an elite consensus and find a majority among the people and in the cantons. What happens in the process of finding a consensus in the political elite is therefore comparable. Lessons about the second step can be instructive for some other cases. 5. Three case studies As outlined in the introduction we have selected three cases to serve for empirical testing of our theoretical assumptions. Our cases fiscal equalisation, education and culture are issues that are set between 1990 and 2006 and which touch upon federal

11 relations. There have been earlier constitutional change attempts in the fields of culture and education prior to 1990, but they are secondary in the context of this study. Before explaining the structure of the narratives and the presentation of the results in the form of a synthetic table, we need to present the federal historical context common to all three cases. This allows for better understanding of the dominance of certain ideas and the behaviour of actors in our three cases. Federal context The main event during the period of investigation was the general revision of the Swiss constitution, the first one after 1848. The revision was intended to streamline the existing constitution. This streamlining was, however, also the occasion to discuss in a more fundamental way the basic principles the federal order in Switzerland should be based upon. Some shifts in the governance of federalism were decided upon. To make a long story short, there were two important principles that were underlined: First, the informal existing cooperative federalism was fixed within the constitution in Article 45: the federal government is obliged to inform the cantons as early as possible about law proposals that concern the cantons and to ask for their statements. The cantons obtained the formal right to participate in policy formulation on the federal level. This was a shift from self rule to shared rule (Fleiner, 2002). Second, its was explicitly mentioned in Article 47 that the federal government had to respect the autonomy of cantons and also that the principle of subsidiarity should hold in federal relations. Both amendments can be seen in the light of a continuing creeping centralisation that had taken place since the 1960s in which the federal government had become co responsible for a large number of policy fields. The revised constitution can therefore be seen as a reconfirmation of the central role cantons play in Swiss federalism, the maintenance of their autonomy, and the protection of their rights by a formalised participation in policy formulation. This has influenced discussions on federal relations in the period. Narrative structure Our case studies that are presented in the appendix start with an overview of the historical background and the definition of what the process is about. The degree of conflict is assessed by three dimensions: efficiency, value cleavages and (re)distribution of money and power. The ideational dimension of the processes are discussed in terms of focal ideas that structure arguing, encompassing frames and the constitutional theory

12 of the actors, meaning their beliefs about what has to be done. We distinguish and analyse three arenas in the processes; idea generation, agreement reaching and ratification arena s. We pay attention to agenda crafting and the role of timing, the use of experts and the inclusion or exclusion of actors. All these elements are used to explain the outcome of acceptance or failure in the ratification arena.

Table 1 Summary overview of three cases of constitutional amendments in Switzerland Variables/ Cases Equalisation Education Culture Intended change Reorganisation of competences between the federation and cantons; new fiscal equalisation. Improved coordination between cantons and the federation in education policy. Constitutional article to define federal competences in culture promotion Perceived need for change Financial crisis in the 1980s and report of federal finance administration report reveal inefficiency of status quo. Shared believe that status quo inefficient because of poor vertical and horizontal coordination. Problems of harmonisation and adaptation to internationalisation and knowledge society Perception that federation s competences do not correspond to social reality. Promotion of culture seen to be too important to be missing from constitution. Initiating factor Parliamentary initiative against centralisation in the 1970s and financial crisis in the 1980s that revealed financial weakness of some cantons. Parliamentary initiative for single educational area in 1997. Expert report (1975) that assesses need for federal culture policy. Political process started by popular initiative in 1981 with reference to expert report. Institutional Procedures Federal finance administration (DF) and Conference of Cantonal Finance Minister (CCFM) mandate economists to find solutions for problems related to federal fiscal system (1994); Mixed cantonal and federal project group (1994); first official consultation procedure (1996); Based on reactions of the consultation, new and larger project group with two issuespecific commissions; Project group submits report to federal government (1999); Second consultation process started, reveals hesitance from many actors; new expert study commissioned. Government sends proposal to parliament (2001), which debates it from 2002 till 2004; referendum in 2004, proposal accepted. Parliamentary initiative 1997; Acceptance and delegation to parliamentary subcommittee (PSC); refusal of the proposal by the conference of cantonal ministers (CCM); Joint working group PSC and CCM; Adoption of proposal by parliamentary commission in 2004; Official consultation procedure; Joint stakeholder convent for consensus crafting (2004); Deliberation in parliament with adjustment of differences between the two chambers and adoption of the final draft (2005). Acceptance in referendum (May 2006). Popular initiative in 1981; government prepares counter proposal after consultation of concerned stakeholders. Double rejection of initiative and counter proposal in vote 1986. Relaunching of process the same year through motion in national council. Administration prepares proposal. Strong agreement in parliament. Rejection of proposal in 1994 vote due to majority of opposing cantons. Successful introduction of culture promotion article as part of general constitutional revision in 1999.

14 Variables/ Cases Equalisation Education Culture Degree of politicisation Redistribution of authority cause for parliamentary initiative in the 1970s; Inefficiency of system responsible for process initiation; Argument of efficiency primes and helps to keep redistribution of authority dimension low. Consensus on need for coordination. Authority redistribution politicized. Consensus on need for adaptation to enhance efficiency. Distribution of money and authority politicized, but taken into account in government s counterproposition. Inclusion of Veto Players at different stages (3 arena s) Exclusion during idea generation to prevent unwrapping of package; Wider inclusion of federal and cantonal actors during agreement reaching. Inclusion of cantons during preparliamentary agreement reaching through joint working group. Only formal inclusion through consultation procedure. Focal ideas and frames Efficiency, new public management, subsidiarity. Efficiency, subsidiarity Subsidiarity; article is declaration of principle, not of action. Expertise Very important for problem definition (1994, 2001) and for reassurance of vetoplayers that project is in line with original intention Experts as moderating element during joint working group. Only present in idea generation arena at a very early stage (1975 report). Shared constitutional theory Economic theory of federalism; Inefficiency of the status quo. Yes, measures to improve efficiency. Introduction of culture promotion competences of the federation within the limits of subsidiarity in the constitution as declaration of principle. Compensation payments Yes. No compensation payments. No compensation payments. Communicative discourse Careful communication strategy based on public opinion survey. Large package of constitutional amendments. Presentation of only this issue on referendum day. No special communication strategy but active engagement of elites. Medium sized package of constitutional amendments. Presentation of only this issue on referendum day. No elite investment in referendum campaign. One single issue of constitutional amendment. Presented together with highly salient issues on referendum day.

6. Discussion What can we learn from the three case studies for the building of an explanatory model of success and failure of constitutional amendments? The specificity of Switzerland is, as we described in the section on the institutional context, that ratification of a constitutional amendment is not just a matter of finding support in parliament as in the case of a representative democracy, but also among the people and cantons. Successful constitution making must therefore find both a consensus in the political elite and among citizens. To be more specific: we think that it, first, needs a large consensus in the political elite before, second, the constitutional amendment may have a chance to find a majority among the population and the cantons. A causal model, which takes direct democracy into account, must therefore be subdivided into a model that is supposed to design necessary and sufficient variables that may create an elite consensus at the level of the political system (the coordinative discourse ) and a model that deals with the communicative discourse or the referendum arena. Modelling the coordinative discourse We have used the problem triangle sketched in the beginning (see figure 1) as a heuristic device in order to classify the kind of problems that may be at the base of constitutional amendments: There are three kinds of problems that political elites must deal with in constitution making: efficiency, distribution, and value problems. Our initial assumption was that the conflict level would be different depending on the kind of problem: if a problem or, in our case, a constitutional amendment is perceived as a pure efficiency problem, then the chance to find a consensus would be very good because discussions would be on a level of problem solving (Scharpf, 1997), i.e. directed to the search for adequate solutions to the problem at hand. Distribution problems (either in terms of authority or money) raise the stakes and can mobilise significant veto potential. The same holds for value problems that, in addition, may have a very unmanageable component because of the absolute character of value cleavages. We are convinced that an elite consensus needs therefore as a necessary condition the following components that are integrated within the following equation:

16 EC = E d v Where EC is elite consensus ; E is efficiency ; d is absence of salient distributive conflicts ; and v is absence of salient value conflicts. In other words, if political entrepreneurs want to make constitutional amendments succeed, they should do everything to make or maintain efficiency as the dominant frame in the discussion and try to downplay imminent distribution and value conflicts. The phrasing of the previous sentence indicates that it often needs a process (in which artistry or agenda crafting will play an important role) to establish the necessary condition for success: efficiency may be at the origins of the constitutional discussion (and in fact this has been the case in all three of our cases) but it may lose momentum, be contested by contradictory constitutional theories or be substituted by distribution and value problems. Distribution and value problems arise most of the time because solutions presented for efficiency problems need a distribution of authority and/or money and touch upon value questions. In fact, distribution or value problems show up in all our three cases and could not be avoided. It is therefore the process that must help to establish the elite consensus. Let us have a look at each necessary variable: Efficiency Efficiency can be the leading dimension at the outset of the constitutional discussion or not. If distributional or value questions start a constitutional initiative, then the level of polarisation will be instantly higher and it will be difficult to construct a large consensus among the elite. To transform distributive or value questions into efficiency questions would, of course, be a way to overcome polarisation. As we do not have such a case in our population, we are not able to suggest possible ways of upgrading efficiency in the coordinative discourse. All our cases have an efficiency question at their base. One finds one accompanying variable, though, that seems to be of importance in making efficiency a positive element for elite consensus and which was also mentioned in the discussion of our causal dimensions, i.e. the existence of only one constitutional theory. By that we mean that there is only one causal story or narrative which is accepted by the political elites and that interprets the problem at hand and can suggest solutions. If there are two or more contradictory constitutional theories, solutions to efficiency itself

17 become a contested issue and efficiency cannot any longer serve as a consensus mechanism. Such a constitutional theory can exist already at the outset of constitutional discussions but, and this is the reason why we discuss it as a process aspect, it can also be searched for and be the outcome of the constitution making process. It can even be deliberately designed. This was the case when the expert report for the reform on equalisation suggested the intimate nexus between competence and equalisation reform on the base of a coherent theory (economic theory of federalism). Experts were actually mandated to present such coherent solutions to avoid inefficiencies of the system. This was not the case with the reform in education and culture. The parliamentary initiative in the case of education suggested clearly an efficiency based account of the problem and suggested solutions that were not contested by any other actor. The granting of additional powers to the federal government in order to overcome inefficiencies in the promotion of culture proposed by the federal administration was not contested as an efficient solution nor were alternatives presented. This means that efficiency as such is not a necessary factor but only in conjunction with the availability of one and only one constitutional theory: EC =(E CT) d v where CT is one constitutional theory. Moreover, one can suggest, on the base of the equalisation case, that, in order to build the dominance of only one constitutional theory, the use of experts is an important factor as might be a certain closure of the idea generation arena. Such a closure makes it easier to develop shared views in epistemic communities. The selected constitutional theory acts then as a filter in the agreement making arena. The more complex the topic is, the more difficult it becomes to develop contesting constitutional theories later on. However, these findings are only built on one case and are not confirmed by the two other cases where the existence of efficiency as the outset dimension and one constitutional theory was a given fact and did not need further artistry. We can only suggest for further studies that, if constitutional theories must be built during the constitution making process, the variables experts ; restricted access to the idea generation arena and complexity of the topic, may be necessary factors to be successful in doing so.

18 Distribution Conflicts In all our cases distribution conflicts could not be avoided during constitutional discussions, even if efficiency was at the base of the discussions. The reason for this is, as explained, that solutions proposed for reasons of efficiency have effects on the existing material or ideational distribution. This makes it necessary to find strategies to downplay such conflicts. We believe on the base of our empirical material that the following equation holds: d = FP I EX where d is absence of distributive conflict ; FP is focal point or equivalent reassurance mechanisms ; I is active inclusion of veto players ; EX is experts. One of the main mechanisms for trust building among actors and for their consent to constitutional solutions are reassurance mechanisms. Actors need to know that they will not become victims of predatory behaviour of other actors and that their standard interests will be protected by future solutions. Focal points are such a mechanism (see above). They are basic ideas or notions that can be used to reveal contract breaking, which can then be sanctioned. In our context subsidiarity has been such a focal point in all three cases. Subsidiarity means in a federal context that the federal government would respect the autonomy of cantons and their prerogatives in executing policy functions. It also meant that the cantons acknowledge the rights of the federal government to act in certain defined areas and circumstances. This principle was therefore meant to avoid interference and to respect the balance of forces. The virulence of this principle, which became a general and overarching notion for federal relations in the 1990s, can be seen in all three cases: It was explicitly inscribed into the mandate of the first project group of the equalisation project and remained a basic principle of reference during discussions; it was time and again mentioned in the discussions on the education project and guided the main decision to leave it to cantons to coordinate themselves before the federal government could come in; and it has been the main principle of reassuring cantons about federal intentions in the promotion of culture: activities of the federal government were explicitly formulated as additional projects and projects that filled in where cantons would or could not act. Reassurance mechanisms can also be concrete agreements on the distribution of resources that are

19 designed to protect the interests of all participants. If agreed upon, they can reduce the salience of distribution conflicts. The active inclusions of veto players is not equivalent to what is a prescribed institutional procedure in Switzerland, i.e. the consultation procedure. The last one can serve to check if a government proposition finds consent among stakeholders and it may also serve as a way to understand technical problems or main points of resistance. The government is not obliged to react. The consultation procedure was used as the only inclusion mechanism in the case of culture and served to confirm that the counterproposal of the government would find a large majority among stakeholders. The active inclusion of veto players means more. It is the often early attempt of the federal administration to include main veto players into the policy formulation process, even before the consultation process begins. It is set up in an interactive way by participation in working groups. This was extensively used in the case of the equalisation agreement and, at a later stage, also in the case of education where cantons were integrated after they opposed openly to a government proposal but still before the official consultation procedure began. The reasons for success are obvious: cantons had not only voice on the federal level but were equal to federal actors in formulating policies or constitutional amendments. Their veto capacity was early taken into account and by participating in the formulation of amendments the danger of resistance at a later stage was significantly reduced. But is it a necessary condition for reducing distributional conflict? It is only under certain circumstances, i.e. in the case that the risk that veto players will use their veto powers is perceived to be as high. This was both the case with equalisation and education. In the case of equalisation, the reform of the general framework, the profound revision of competences and the manifold questions linked to revenue distribution made veto s very probable. In education, integration occurred after resistance was already openly pronounced. The contrasting case is culture: Here we do not find an active inclusion of actors because the value conflict dimension was since the beginning dampened by the general acceptance of the focal point subsidiarity and the distributive conflict was only represented by a minority in parliament. Salience of the conflict in culture remained in general very low. Therefore, active inclusion cannot be a necessary condition as such. It becomes a necessary condition if there is a high risk of the use of veto potential of actors.

20 The same conclusion can be drawn with regard to the next variable, experts. We can indeed see in both empirical cases with high veto potential that experts have had an important moderator function: in the case of equalisation, an expert report was used to overcome resistance and hesitance by a number of veto players apparently with success; and in the case of education, experts were used as mediators between the federal government and the cantons. They were not used, however, in the case of culture, which confirms our previous conclusion: both active integration of veto players and experts is conditional upon the veto potential of issues. Reassurance mechanisms by contrast are unconditional. There is still one other modification to be made. The formula stipulated above holds without any doubt for those cases where authority distribution is the topic. If, however, it is not authority distribution or not only distribution but also distribution of money, an additional condition may hold: there must be room for compensation payments. We deduce this from our case on equalisation, which is the only case with revenue distribution. The more it became transparent what the new distribution of competences and rules for equalisation meant, the more bargaining entered into the coordinative discourse of elites. If on that occasion there had been no room for side payments, compensation payments etc., acceptance of the law project would probably have failed. If therefore money distribution is an issue, the disposal of compensation payments seems to be necessary. But again, we can only suggest this as an important variable but not confirm it as a necessary cause as it is based on only one case. In sum, artistry in downplaying distributional conflicts is the active use and promotion of focal points and other reassurance mechanisms as well as the active inclusion of vetoplayers and experts in the case there is a high veto potential and of compensation payments in the case of money distribution. A case with low veto potential and authority distribution would therefore completely rely on focal points or reassurance mechanisms. This indeed was the case with culture. Value conflicts Conclusions with regard to value conflicts are not radically different from the ones drawn on distribution conflicts. Value conflicts have played implicitly a role in all three of our cases as culture, education, and the topic of compentence distribution and euqalisation could in principle be linked to regional identity. The potential to politicise

21 these constitutional items in terms of value conflicts was there. Other value conflicts could also emerge as the social equity question in the case of equalisation demonstrates or the progressive traditional divide, which showed up in the referendum arena of the culture project. With regard to contingent value conflicts that may come up unexpectedly as in the case of the two issues last mentioned, it is difficult to define any general process variables that could help to downplay the conflict. The progressive traditional divide was a total surprise for political elites and no precautionary action was taken. The social equity question was elegantly dealt with by using the territorial dimension against the left right dimension in the party system. But this is something, which can certainly not be generalised. We can, however, say something about the regional identity dimension, as it played a potential role of conflict in all three cases. As in the case of distributional conflicts it seems that a focal point like subsidiarity has been the most important tool to prevent a higher salience of this conflict. A value conflict in the sensible area of culture could for example be avoided because of the clear announcement of the federal government to respect the subsidiarity principle. The settlements in education respected, as already said, also the subsidiarity principle. In the case of competence assignment and equalisation the question of regional identity came never on the political agenda, in part certainly because all activities by the federal government expressed the respect with regard to this principle since the beginning. The other variables we discussed in the case of distributional conflicts active inclusion, experts, and compensation payments seem not to play a significant role in value conflicts linked to regional identity. Arguing or bargaining do obviously not make much sense if there is polarisation on the value dimension as it is not possible to build compromises or pay compensations in this case. The winning strategy is reassurance, i.e. to guarantee to minorities that their identity is respected in the constitutional process. Focal points and other reassuring mechanisms is therefore the sufficient mechanism for downplaying value conflicts.

22 Modelling the communicative discourse The creation of an elite consensus is important but in a referendum democracy it is not sufficient for successful constitutional change. In addition, it needs a majority in the population and, in the case of an obligatory referendum, also a majority of cantons (14 out of 26). We contend that there are four necessary factors that in causal conjunction are responsible for successful constitutional change: a large elite consensus; the complexity of the issue presented (packages); the presentation of only one issue on referendum day; and an active engagement of elites in the mobilisation of the people. RS = EC P N EE Where RS is referendum success ; EC is large elite consensus ; P is complex package ; N is non issue linkage ; and EE is elite engagement. (a) The construction of a large consensus in the political elite is without any doubt a necessary condition for successful constitutional change. If a constitutional amendment is seriously contested in parliament and for example approved by a small majority only, a serious contestation and mobilisation against the amendment can be expected in the ratification arena and the chances are high that the amendment will not pass. A large consensus in the political elite is, however, not a sufficient condition as our cases demonstrate. The contrasting case is the amendment on culture that was refused despite of a very large elite consensus. We think therefore that there are three other necessary factors that must be added in order to make constitutional amendments successful: (b) The more complex the issue is people must decide upon, i.e. the more constitutional articles must be changed that are presented in the form of a package, the more difficult it is for people to reject the proposition of the political elite. The sunk costs are in this case very high. In the cases of education and equalisation the amendments were such packages including complex and interrelated topics needing the change of several constitutional articles. The case of culture confirms the hypothesis: it was only one article that had to be changed with no overt complexity entailed. Untying a package in a popular referendum is technically not possible. Consequently this leaves the electorate only the possibility of taking or leaving the deal as a whole.