Political Parties as Multi-Level Organizations

Kris Deschouwer. Handbook of Party Politics. Editor: Richard S Katz & William Crotty. 2006. Sage Publishing.

The Obvious National Context

One of the major assumptions of much of the analysis of party politics is that it is situated in the context of national states. The national state is therefore normally seen as the primary institutional context shaping party politics. The direct and evident association between parties and the state is the result of the historical coincidence of state formation and the development of mass politics. The formation of parties organizing and shaping public opinion occurred within the territorial limits of the national states and often even along cleavage lines that were at the heart of the state formation process itself, like the tensions between church and state and between center and periphery (Lipset and Rokkan, 1967; Flora et al., 1999). Crucial notions such as citizenship, and its concrete derivatives such as voting rights and democratic representation, were defined in this same environment of the national state.

Scholars of political parties have analysed these origins and further developments of political parties, and have therefore taken the national parties as their most obvious unit of analysis. And if a system of parties was being studied, this referred to the interactions of the party units inside one single statewide political system. Of course not all modern states are unitary states, and parties have developed and been analysed in federal states as well. This has, however, mainly been done by looking at them state by state. The comparative literature on political parties in federal states is rather scarce (Chandler, 1987; Scharpf, 1995). One of the reasons is probably that federal states are indeed different from unitary states, but do not all function in the same way. And these internal differences might be crucial for the understanding of the political parties. Furthermore, when parties have been analysed in federal settings, the assumptions still were that the central state was the most important center of decision-making and that the central organization of the political parties was the core of the parties, the reference point for analysing their internal organizations. Regional parties have received attention in the very specific literature on regionalism (De Winter and Türsan, 1998), but this has also paid little attention to the specificities of the institutional context in which they function.

To assume that parties function either in a unitary or in a fairly centralized and hierarchical federal state context is no longer tenable. Many states have been going through rapid and sometimes spectacular processes of devolution and decentralization: Spain, Belgium, Italy, the UK, France. This has led to increased attention for the functioning of party politics in these new settings (Jeffery and Hough, 2003) and to the search for a proper comparative language to deal with them (Deschouwer, 2000b, 2003; Hopkin, 2003; Thorlakson, 2001). The gradual integration of the European Union and the direct election of the European Parliament have also created interest in the dynamics of party competition in complex and multi-level systems (Reif and Schmitt, 1980; Reif, 1984; Andeweg, 1995). The very peculiar polity that is the (temporary) result of European integration and the very varying and mostly asymmetric state institutions that have recently been put into place, have also made clear that the notion of ‘federalism’ has become either too limited to encompass the meaning of the new political institutions, or—if all the new variants are included—too broad for good analysis. The conclusion is that we need to question and problematize the institutional context in which political parties function. This context must be one of the crucial variables—because it varies indeed—related to the strategic and organizational choices of political parties.

 Table 24.1 A typology of parties in multi-level systems


Table 24.1 A typology of parties in multi-level systems

In this chapter we will focus on political parties in multi-level systems. In the first place we will develop a classification of parties that allows us to identify specific characteristics of parties in multi-level systems. Then we will explicitly bring in the institutional context, to show how it directly affects the position of parties in the system and therefore their functioning. We will decompose the broad notion of federalism or multi-level system into three dimensions: formal institutions, electoral rules and cycles, and societal heterogeneity.

A Typology of Multi-Level Party Organizations

Parties in multi-level systems face particular problems that are the direct consequence of the organization of the political system. In order to analyse properly the functioning of these parties, we need in the very first place to identify their specificity. We need to identify variations between parties that are typical of a multi-level system. Attempts to do so have been very much inspired by the Canadian party system, with its separation between federal and provincial parties and party systems (Dyck, 1991; Smiley, 1987; Thorlakson, 2001). Two dimensions of variation have been identified, but have to be kept separate because they are to a large degree independent of each other.

The first dimension is the presence of a party at the different levels of the political system. The second refers to its territorial pervasiveness. That is a different dimension indeed, because the presence of more than one level on the same territory is exactly the typical feature of a multi-level system. Both dimensions can be combined into a typology with nine logical positions, as presented in Table 24.1. For the sake of clarity and of parsimony we reason within the logic of a political system with two autonomous levels of decision-making. We use the term ‘regional’ to refer to the lower level and ‘national’ to refer to the higher level. Most of the examples will be drawn from federal or decentralized states, but we will refer once in a while also to the European Union.

Parties in multi-level systems are confronted with the choice of participating in elections at only one or at both levels of the system. That leads logically to three types of parties. The first is a party that participates at the regional level only. It is thus typically specific to one single region in the multi-level system. The Parti Québecois in Canada is a good example of this. Another example is the Partei Rechtsstaatlicher Offensive in the German Land of Hamburg. One might say that this type of party does not require special theoretical attention. It is indeed only competing in one single, and single-level, system, and can therefore be understood with the same theoretical tools as those used for the analysis of national political parties in their national party system. This is, however, only true to a limited extent. Indeed, the fact that this party functions in a multi-level system can have quite important consequences for the way in which it behaves, for its political opponents will be engaged in the more complex competition at more than one level.

The second type is the party that participates at the federal level only. This is a rather exceptional situation in federal systems, but the category is certainly not empty. The Canadian Bloc Québecois, for instance, participates only in federal elections. If we look at the European Union as a multi-level system, more examples can be found. Parties that participate in European elections only, and not in national elections, can be found in several countries. The specific meaning of the European elections, and the opportunity they offer to voice Eurosceptic attitudes, make this level the ideal one for these parties to operate at (Taggart, 1998). They could of course also defend this Eurosceptic view in national elections, but there are a few good reasons not to do so. One of these is the ‘division of labour’ between national and European elections, where the European dimension is avoided in national elections (Mair, 2000). Another reason for not running at the national level can also be the electoral system. If in France the Rassemblement pour la France of Charles Pasqua engages only at the European level, it can do so because the electoral system is proportional at that level and because the bipolar structure of the French national political competition—kept in place by the majoritarian electoral system—is very weak in European elections.

The third type groups the parties that participate both in regional and in national elections. This category is much more populated than the previous ones. Many parties in multi-level systems do indeed participate at both levels. Most of the regionalist parties in the autonomous communities in Spain for instance, are present in the regional elections (where some of them try to win office) and in the elections for the Cortes in Madrid. The major political parties in most multi-level systems also compete in both elections.

The second dimension is the territorial pervasiveness of the parties. This is actually a continuum, varying between parties being active in only one of the regions to parties covering the whole territory. In Table 24.1 we have divided this continuum into three broad categories. The first—parties covering only one region—are regional parties or parties that are typical of one single region in the system. The Partei Rechtsstaatlicher Offensive only covers Hamburg. The Catalan Convergencia i Unio and all the other regionalist parties of Spain only cover their own region. The same goes for the Scottish Nationalist Party, Plaid Cymru, etc. The German CSU or the Parti Québecois and the Bloc Québecois are other examples. All the Belgian political parties also limit their activities to one of the two main language communities of the country. This illustrates also that the territorial pervasiveness differs from presence at one or two levels. Parties covering only one region can indeed choose to participate only at the regional level, only at the federal level, or in both levels.

The second category consists of parties that cover more than one region, but not the complete territory of the multi-level system. Some parties cover a few regions, such as the Lega Nord in Italy or the PDS in Germany. Both also participate in elections both in their regions and at the federal level. Germany and Spain offer interesting examples of parties that are almost complete, but are absent in just one region. That is the case for the German CDU, leaving Bavaria to the CSU. It is also the case for the Spanish PSOE, leaving Catalonia to the Catalan PSC.

Finally there are the pervasive or complete parties that cover all regions. This category is populated with the American parties, the major Swiss parties, the Spanish Partido Popular, the German SPD, the Austrian parties, the Australian parties, etc.

As we already said, not all categories are empirically very densely populated. Most parties have to be placed in the right-hand column of parties being present at both levels, with territorial pervasiveness as a meaningful further classification. Categories 1 and 4 are less populated but contain some cases, mainly from Canada and from member state party systems in the European Union. The relative emptiness of categories 5 and 6 (parties present only in federal elections but in more than one region) actually illustrates a problem of definition. If we want to classify parties, we need to identify the unit of analysis that will be so called. Especially in multi-level systems, the identification of the unit party can be very tricky. The Canadian parties—always difficult to place -are a nice illustration of this. Dyck (1991) has labelled them ‘truncated parties’ (see also Wolinetz and Carty 2006) because the provincial organization and the federal organization are so separated from each other, with separate membership and separate finance, that they can hardly be considered to be one single organization. Still doing so reflects the implicit assumption that there is a single organization party of which the provincial units are a part. If the separation goes as far as in the Canadian case, one might as well choose to define the parties as two different units: for instance, a Liberal Party participating at the national level only, and provincial parties participating at the provincial level only. For both we can then add the distinction between presence at only one or at both levels. What used to be studied as internal relations of one party organization then becomes the analysis of relations—both horizontal and vertical—between parties of the same ideological family within a multi-level political system (Deschouwer, 2003).

The two dimensions of variation presented in Table 24.1 are both relevant for the party organization, but in different ways. The major issue is that of vertical integration. It occurs when a party is present at more than one level. Parties present at only one level have only one organization and a single strategy to develop. That is the case for parties of type 1 and 4 in Table 24.1. Adding a second level introduces a potential tension in the organization. There must be some place in the organization where the two levels can be coordinated. The degree to which the party allows internal division and eventually formal organizational divisions to function at both levels is a major research question for this type of party.

If a party is not territorially limited to one region, the problem of vertical integration is (potentially) of a different nature. Not only is there the need to coordinate between the two levels, but also the need to coordinate and control for horizontal variation across the regions. In other words, the party needs to organize in a way that allows it to deal with the territorially varying problems of vertical integration. It is clear that the type of society in which the party functions (the territorial heterogeneity) is one of the major causes of potential difficulties in integrating the varying demands.

If vertical integration is high, the different levels and the varying territories have a limited degree of autonomy. In that case the hierarchical lines in the party are clearly going from the national to the regional level. If integration is low, the regional organizations of the party have some freedom to make their own decisions. This freedom does not need to be uniform. In pervasive parties the regional levels can be very different and therefore enjoy a different level of autonomy.

The degree of autonomy of regional-level organizations is not a one-dimensional phenomenon. Several indicators can reveal varying forms and degrees of autonomy. A first set of indicators refers to the party organization and thus to the autonomy of the regional branch. The membership structure, for instance, can be very revealing. In the German and Austrian parties, the members join at the regional level, and this membership automatically implies membership of the national party. In the Swiss parties, the cantonal level is clearly the most important in this respect. Members join at the cantonal level, and this is the only membership possible. The membership of the national parties is indirect: the cantonal parties join, not the individual members. In Canada, some of the parties (such as the Liberal Party and the Progressive Conservative Party) are so loosely coupled that they have actually two different membership organizations at the regional level: one for the national party and one for the regional party (Thorlakson, 2001).

The recruitment of political personnel is another indicator of regional autonomy and also often the stakes of conflicts between the levels. The British Labour Party has been trying to control leadership and candidate selection at the regional level in Scotland, Wales and Greater London, but has not been able to keep the same degree of control of these areas as for the rest of the country (Hopkin, 2003). Strong regional leaders can—in systems with real regional political autonomy—build their legitimacy on strong regional electoral results and put pressure on the national leadership. Political recruitment in multi-level parties is often a bottom-up affair, where strong regional leaders (eventually leading the regional government) make their way up to the national leadership.

Membership, recruitment of leadership and of candidates for elections, but also, for instance, financial autonomy and control over the lower levels of the party organization (the local sections) are all indicators that allow one to see the extent to which the two levels are separated and whether the hierarchical lines go from top to bottom or vice versa. A guaranteed presence of the regional sections in national party decision-making is another useful indicator.

Another set of indicators focus on party strategies. Parties in multi-level systems operate in varying political conditions and can need very distinctive strategies for each of these. These strategies include the decision to participate in elections, campaign strategy, formulation of policy proposals and coalition strategy. For the latter, the developments in the German SPD are interesting. Reunification and the presence of the PDS in the five new Länder have created tensions. Traditionally the national party organization kept a fairly high control over the coalition strategies of the regional units, but it has been confronted with the demands of the Eastern regional units to decide more freely on the choice of the coalition partners, including the PDS (Jesse, 1997).

A good deal of variation can be seen here, both within and between multi-level systems. The Australian Labor Party has gradually become more and more integrated, and has strong control over the activities of the regional branches. It can even expel a regional branch from the party. The Australian National Party, on the other hand, remains much more decentralized and has a more confederal type of national organization (Thorlakson, 2001). The Swiss parties are again an example of a high degree of regional autonomy, although—as in Australia—the Socialist Party is much more integrated. Cantonal parties regularly defend positions different than the national party on referendum issues (Sciarini and Hug, 1999). When members of the Swiss federal government are to be elected, both the national party organization and some cantonal organizations often propose and defend different candidates. The American political parties are another example of very loosely coupled parties (Katz and Kolodny 1994).

Explaining the Variation: The Institutional Context

The degree of vertical integration of a party or the autonomy of its regional branches can vary widely. This variation can occur both between countries and within countries. It is in the first place related to a party’s position in the system, i.e. to its place in the classification presented above. This position, however, is the result of developments and eventually deliberate choices of the parties operating in a very specific institutional environment. Both its position in the system and the way in which it deals with vertical integration and horizontal variation are related to that environment. That is why a proper analysis of parties in multilevel systems needs to bring in explicitly the characteristics of the institutional environment. It can be divided into three dimensions: formal institutions, electoral systems and cycles, and societal heterogeneity.

The Formal Institutions

Analysing parties in federal settings has seldom been done in a comparative way. The main reason is that federalism is too broad a concept. It needs to be broken down into concrete characteristics that can be linked to party politics. Actually, only one feature has been put forward and identified as crucial for party politics: the distinction between dual and cooperative federalism (Chandler, 1987; Scharpf, 1995). It is indeed important. If the federal (or multi-level) system displays a neat division of competencies between the levels, the relations between the two levels of policy-making are limited. That allows and pushes the parties to have fairly autonomous regional branches. Regional autonomy allows the most suitable strategies for that level to be chosen without interfering with the activities of the party at the national level (see also Thorlakson, 2001; Deschouwer; 2000b, 2003).

If the levels of policy-making are interconnected, the relations between the regional and the national branches of the parties take on a different nature. This interconnectedness can be due to the fact that the allocation of competencies explicitly allows for mixed areas, in which both levels need to move together. That is the case in Germany. It can also be the consequence of a functional logic of the multi-level system (as opposed to a jurisdictional logic), in which the higher level produces general framework laws that need to be implemented by the lower level. That is, for instance, very much the case for the European Union. A third source of strong linkage between the levels is the formal presence of the regional level in national decision-making. The German Bundesrat organizes this presence, while in Canada and in Australia it is formalized in conferences of the prime ministers (and in the EU in the Council of Ministers). The consequences for the functioning of parties are evident: they become very deeply involved in intergovernmental relationships. Intergovernmental politics becomes party politics and vice versa (Lehmbruch, 1976; Rydon, 1988; Jeffery 1999; Hadley et al., 1989). The regional branches then cannot be too autonomous. Regional policy-making and regional elections become relevant for federal policy-making and will be framed in these terms.

Another dimension of the formal institutions of multi-level systems that is closely connected to the previous one is the degree of autonomy of the regions. Indeed, if the degree of autonomy is low, the logic of the distribution of the competencies is not likely to play a significant role. It does, however, if (some of) the regions have the real control over a number of important policy domains. This is of course a matter of degree, for which it is not easy to define a clear cut-off point. In unitary states, there can certainly be some degree of decentralization, but since it does not go very far and does not give the lower level a sufficient degree of autonomy, the analysis of party politics does not need to bring in explicitly the institutional environment as a variable shaping party organizations and strategies. The need for a multilevel game and for the parties to incorporate it into their organization only emerges in a system where the regional level has at least some degree of sovereignty.

The other extreme of regional autonomy can have far-reaching effects on the political parties. If the autonomy of the lower level goes very far, as in Switzerland, the USA or Belgium, the regional branches of the parties become the core of the party, and the higher level—for parties crossing the levels—becomes a loose association of the regional branches. The lines of command in the party—if we accept that the higher level is still considered to be a party—then go from bottom to top. In such a situation, it is not unlikely that the parties organize separately for regional and national matters, as in Canada, or simply fall apart into separate regional parties, as in Belgium (Deschouwer, 1994).

The degree of autonomy and the type of distribution of competence are two variables broad enough but still relevant to be used in comparative analysis of party organizations in multi-level systems. They can and must also be used for comparative analysis within multilevel systems, when these systems are asymmetric. In the examples that we have given so far, we generally referred to the parties of one or more countries to illustrate how the institutional context affects party life. But party life itself becomes asymmetric when the system allows for varying degrees of autonomy between regions and varying logics of the way in which competencies are distributed between national and regional level. The British system after devolution and the Spanish system are good cases for looking at the way in which parties adapt to this institutional variation. Degrees of vertical integration and regional branch autonomy vary then between regions.

Electoral Systems and Cycles

If assemblies and eventually also executives have to be elected at different levels, the techniques used to do so can differ. This might again also be the case in unitary systems, where lower tiers (provinces, local municipalities) have elected bodies. But in systems where the different levels have real powers, variations in the way in which elections are organized can have quite far-reaching consequences. The number of seats available per region is normally higher in regional than in national elections. In majoritarian systems this does not affect party strategies, but it clearly does in proportional systems. The higher number of seats at the regional level leads in general to lower thresholds of representation, and this can influence a party’s decision to participate at the level with the lower threshold, and its decision to go alone or in association with another party (Lutz, 1998; Lancaster, 1999).

This effect is very visible in European elections, where in 1999 all countries adopted a proportional system. For some parties in France and in the UK, the European level suddenly offered better possibilities for representation than the national elections. We have already mentioned the Rassemblement pour la France. In the UK, the Green Party and the UK Independence Party were able to gain European seats, while they have no chance at all of gaining representation at the national level. Within the UK, the additional vote system for Scottish Parliament and Welsh Assembly elections also offers a different environment at the regional level.

Much more important than the electoral systems and formulae is the timing of elections. The idea that elections at one level are in one way or another linked to the elections at the other level has been developed mainly in the German context (Dinkel, 1977), with its relatively high number of regions organizing elections at different times. The direct election of the European Parliament from 1979 on and its analysis by Reif and Schmitt (1980) has led to the notion of ‘second-order elections’. The idea is that this election of the members of the European Parliament by country cannot be seen as a European election. It is a series of national elections with the national parties competing, but with less at stake. The European election is framed in national terms but the outcome has no impact on it, as a first-order election would have. Reif (1984) has refined the analysis of the interaction between levels by drawing attention to the position of the European elections in the national electoral cycle. European elections are not necessarily ‘mid-term’ elections, but can be earlier or later in the electoral cycle, and therefore have some real effects on the politics in the specific country. The notion of second-order elections has—parallel with the processes of devolution and decentralization—been used to analyse the relationship between regional and national elections in multi-level systems in general (Heath et al., 1999; Abedi and Siaroff, 1999; Jeffery and Hough, 2003; Pallarés and Keating, 2003; Detterbeck and Renzsch, 2003).

In fact, there are two different dimensions of electoral timing that need to be taken into account: vertical and horizontal simultaneity. Vertical simultaneity is the coincidence of elections at two levels: a regional election taking place on the same day as the national election. This vertical simultaneity reduces the autonomy of the regional level. The statewide electoral stakes overshadow the regional stakes. For the regional branches of the parties also participating in the national election, the selection of candidates and of campaign strategies and themes will need to be done in close cooperation with the national party level (Versmessen, 1995; Deschouwer, 2000a). That is, however, only valid if the national level of the party is the core level, i.e. if the national party is a vertically integrated party and not a loose association of the regional branches. That latter is the case in European elections, and therefore the simultaneity of parliamentary elections in a member state and elections of representatives in the European Parliament leads to the devaluation of the European elections. Here the national parties are the core level, and not the European party federations that hardly interfere in strategic choices of the national parties.

If the elections at two levels are not organized on the same day, the relation between the two electoral cycles is the crucial contextual condition for the functioning of the parties. Obviously the temporal disconnection offers more opportunities for the regional branches to engage freely in the regional political competition. A regional election organized close to a national election, however, again reduces regional autonomy. A regional election immediately after a national election can easily be read as a confirmation of the national election. Depending on the results of the latter, a regional party branch might try to surf on the winning wave and refer explicitly to the dynamics at the national level, or on the contrary try to make clear that the regional issues are different. In both cases, the connection is explicitly made, and the national party organization will be inclined to monitor the regional elections closely.

If the regional election comes very late in the national cycle and is organized just before the next national election, the chances of being ‘pulled up’ to the higher level are even greater. It is then difficult to avoid seeing the regional election as the final test, and national parties competing in the upcoming national election will make sure that the regional election is organized according to the needs of the national party. A very nice illustration of this mechanism was the election in Lower Saxony in 1998, bringing Gerhard Schröder to the position of prime minister of the region, but in the very first place confirming that he would lead the SPD in the national elections.

If the electoral cycles are disconnected, the position of the regional election in the national cycle is not necessarily fixed. This means that the regional specificity of the election and thus the potential autonomy of the regional branch can vary over time. Party politics in multi-level systems is definitely not a static affair.

The second dimension of electoral timing is horizontal simultaneity: the organization of multiple (or all) regional elections on the same day. Here again simultaneity reduces autonomy. The mechanism is obvious: the results of regional elections on the complete territory of the multi-level polity can be aggregated and thus read within the national frame of competition. In May 2000 most of the Italian regions went to the polls to elect regional parliaments and regional prime ministers. The national opposition leader, Silvio Berlusconi, framed the election as a test of the popularity of the opposition. The results were disappointing for the governing Ulivo coalition, and the national prime minister, Massimo D’Alema, resigned.

Regional elections in Australia, Canada, Austria, Germany or Switzerland are all held on different days. In this condition the regional party branches—depending also on vertical simultaneity and on the type of distribution of competencies and degree of regional autonomy -can be and must be fairly autonomous, and the national party needs to be able to manage the high degree of territorial diversity. In the USA, elections are organized simultaneously, but the other elements of the institutional context do not force the regional parties to stick to a common national strategy and framing of the elections. The different aspects of the institutional environment clearly interact with each other, and that is why one cannot simply refer to a general notion such as federalism or decentralization to grasp the way in which the institutional context shapes parties in a multi-level system.

As with the degree of autonomy of the regions, the logic of electoral timing can differ between regions of the same country. In Italy, regions with a special statute do not have regional elections on the same day as other regions. In Spain, a number of ‘historical’ regions have an electoral timing of their own, while all the others organize their regional elections on the same day. The granting of a different timing to the historical regions illustrates a willingness to give these regions a special and more autonomous statute.

Societal Heterogeneity

Asymmetry has already been raised several times as an element increasing the complexity of a political system and the analysis of the functioning of its political parties, because it is a crucial element that needs to be taken explicitly into account. Asymmetry in multi-level systems has a double meaning (Watts, 1999). It can refer to the asymmetric division of competencies to the regions, i.e. to the fact that regions of the same national system are not institutionally equal. That is what we have taken into account so far. However, we have also implicitly referred to the second one: the territorial variation and heterogeneity of the society. Many of the more recent processes of devolution giving birth to multi-level systems, such as Spain’s autonomous communities, Belgium’s language communities or the UK’s countries, have been attempts to take into account the variations in territorial identity within national states. In these recent examples, societal heterogeneity has been translated into asymmetric institutions. Yet in multi-level systems with symmetric formal institutions the societal heterogeneity can also be a crucial factor shaping the regional dynamics of party politics and thus both the vertical and horizontal relations between party branches. The main reason is the dissimilarity of the party landscape amongst regions and between regions and the federal level.

An ‘index of dissimilarity’ has been developed (originally with reference to Canada) to measure the differences between the regional and national party landscapes (Johnston, 1980; Abedi and Siaroff, 1999). The index compares the results of the national elections in a region with the results of the nearest regional election. The index thus measures the proportion of the voters who voted differently in the two elections. In some cases it can be extremely low—for example, 3% in Baden-Württemburg in 1972 or 4% in Extramadura in 1989. But it can also reach very high levels -26% in Catalonia in 1982 or even 64% in British Columbia in 1974 (Jeffery and Hough, 2003: 208).

There are, however, two organizational challenges for parties in multi-level systems. There is indeed the difference between the national and the regional competition, but the variation among regions needs to be managed as well. If all regions have a high index of dissimilarity or if the index varies strongly among regions, the more pervasive parties have to accommodate to this territorial diversity. Territorial diversity means the existence of different party systems in the different regions and thus varying structures of party competition calling for different strategies and thus for more autonomy for the regional level of the parties (or for the regional branch of the party in the region that displays a deviant structure of competition). Different party systems and structures of competition also ask for more autonomy in the coalition behaviour of the regional branches. We have in this respect already referred to the varying relations between the SPD and the former communist PDS in the five new regions of the German federation.

Conclusion

Political parties in multi-level systems deserve special attention. The institutional context in which they function has to be brought explicitly into the analytic models, because they face very peculiar challenges. Figure 24.1 summarizes this specific analytic logic for the study of parties in multi-level systems. In the first place we need a classification of parties that takes the position of the party in the system into account. The presence at only one or at both levels is the first dimension of this typology its presence across regions or its territorial pervasiveness is the second dimension.

Figure 24.1 Framework for the analysis of parties in multi-level systems

Figure 24.1 Framework for the analysis of parties in multi-level systems

As far as the party organization is concerned, a party in a multi-level system is confronted with two specific organizational problems. The first is vertical integration, the linking of the activities and strategies at two different levels. The second—related to it in a way that depends on the party’s position in the system—is the managing of territorial variation between the regions in which the party participates in regional politics, national politics or both levels at the same time.

In the third place, the analysis of parties in multi-level systems must explicitly problematize the type of system in which the party functions. Classic notions distinguishing between unitary and federal states cannot suffice in a world where the variation between systems has become impressive. Five crucial variables of a multi-level system have been identified. The first is the distinction between dual federalism (neatly separated competencies between the levels) and cooperative federalism (both levels obliged to coordinate their actions). The second is the degree of regional autonomy, which can vary between states but also and increasingly within states. The third is the variation between electoral systems used at the national and the regional level. The fourth is the interconnection of the electoral cycles at the regional and national level. And the fifth is the degree of territorial heterogeneity of the society in which (especially more pervasive) parties have to function.The life of parties in a multi-level system is not a static affair and is not easy to capture in a few straightforward and unambiguous categories and propositions. Parties in multi-level systems live in, adapt to and sometimes try to adapt the complex institutional context in which they function. Giving a straightforward answer to the question how parties organize in multi-level systems is therefore not easy. It depends.