1 I Comparative Federalism and Regionalism in Western Europe: a Conceptual Overview 1. Federalism and federation A comparative study on federalism and regionalism in Western Europe cannot bypass the thorny issue of defining both terms. I shall take it as the starting point of my analysis. As is the case for many general political concepts such as democracy, globalization, or legitimacy, scholars have attributed different meanings to federalism and regionalism. In this chapter, I do not seek to impose my own definition, but shall look for a common denominator derived from a set of several authoritative definitions. Subsequently I will seek to apply them in a consistent way throughout this book so that readers have no doubt as to what is meant whenever 'federal' or 'regionalized' are used as adjectives. I shall also clarify in what sense federalism and federation are different from 'multilevel governance', and how federal (as an adjective) is used differently here from its use by an influential group of (political) economists in their comparative studies of fiscal federalism. Federalism, unlike federation, is in the first place an ideological or normative concept. It reflects a way of living which combines a degree of 'self and 'shared' rule (Elazar 1987: 5). Defined in such broad terms, many attributes of political life, even daily life, can be considered as federal (King 1982: 20). Political systems in which citizens share in the governance of a common system, but retain a certain degree of autonomy within any type of subsystem are federal. Federalism reflects 'unity in diversity' and as such it has been used as a desirable principle of governance for many political systems, including the EU (Burgess 2000). Defined accordingly, federalism is a very broad concept with a strong normative connotation but also with a very limited analytical value. It is difficult to conceive of any contemporary democracy which does not engrain some sort of federalism. Two examples make this clear. For instance, the Scandinavian countries with their strongly developed local tiers of government should qualify as federal. Or consider the following example. In his early writings the Dutch-American political scientist Arend Lijphart developed the concept of consociationalism. Consociationalism describes a mode of governance that has enabled plural societies to survive, despite the presence of deep-seated linguistic, ethnic or religious cleavages (Lijphart 1968). In Lijphart's view the survival of consociational democracies, such as Belgium, the Netherlands and Switzerland, rested upon a number of institutional attributes. In general, they combined a form of power-sharing between representatives of the significant segments (religions, ethnicities, language groups) which compound the state with a form of segmental autonomy. Hence, each of the segments has some freedom to take certain decisions without interference from the others. The 'Good Friday Agreement', which created a post-devolution institutional settlement for Northern Ireland, is one of the most recent examples of a similar power-sharing consociational settlement! Yet, if federalism is simply defined as a combination of self-rule and shared-rule then all consociational regimes should be perceived as federal as they institutionalize a combination of both principles. Most scholars of comparative politics would disagree with such an observation - including Arend Lijphart (Lijphart 1979). In Chapter 7, I will indeed clarify that not all federal states are consociational, just as not all consociational regimes are federal. Consociational regimes can be unitary (at some point in their history, the Netherlands or South Africa were consociational); federal regimes may not be consociational (for instance, the USA federation). When thinking of a federal state, comparativists usually keep specific institutional attributes in mind. These are generally missing from a normative interpretative framework of federalism. To sharpen the analytical value of our comparative analysis, from this point forward, I shall use the adjective 'federal' to refer to a set of institutional characteristics which turn a state into a federation. Hence, when speaking of a federal centre, or a federal state, I mean that the center has all the attributes of a centre in a federation, or more generally that the state has all the attributes that would comply with the features of a federation. But the problem does not stop here. Scholars of comparative federalism disagree on the institutional requirements of a federation." Let me review some of the authoritative definitions. In his influential book on federalism, Daniel J. Elazar defines a federation as a 'a polity compounded of strong constituent entities and a strong general government/each possessing powers delegated to it by the people ancl empowered to deal directly with the citizenry in the exercise of those powers' (Elazar 1987: 7). Elazar's definition implies that, in a federation, the federal and regional levels of government should have a direct link with the people. Consequently, representatives at the regional levels can act without the authorization of representatives at the federal level or vice versa. Conversely, representatives in the federal echelon can take binding decisions for which they do not require the approval of the regions first. However, Elazar's definition does not clarify whether or not the constituent entities must be territorial in character. He does not specify the minimum set of powers which the constituent or general governments must possess to make -them sufficiently 'strong'. All in all, a state should only comply with a limited number of requirements in order to qualify as federal. This helps to explain why of all the comparable scholars, Elazar identified the largest number of federations in the world. The opposite is true for Kenneth C. Whearei whose Federal Government remains one of the most influential works on federalism to date, although nearly six decades have elapsed since its first publication (Wheare 1946). For i KeimethJVh^are, a federation is a political system which applies the 'federal j principle'. BythTtederalprinciple,Jie^meaiis^ffiemet^ : so thatThe~^eneraT~and"Tegional governments are each lwth"in-a-sphere co^OTdmalFaTTdTnaependent' (WheirT:H763rT0)"rSpecify^ ; gOTenaments asT^' / First, the entities that compound the lower order of government should be predominantly though not necessarily exclusively territorial in character. Territoriality is only implicitly implied by Watts's third characteristic. Adding it as a requirement allows us to distinguish federations from some of the consociational polities which allocate autonomy to minority groups such as religious or language communities who live dispersed throughout a territory (Lijphart 1979). Second, the central andjegional orders of governmentsfowM have a democratic state structure. The fede^l^nd-regiOTiaTlegislators musTbe directly elected in a free and open election process. Surprisingly, many scholars of federalism leave out democracy as a .prerequisite of federation. Yet, the notion of 'self-rule' must imply that citizens ought to have a right to elect their public representatives in accordance with their individual political preferences. Consequently, differently coloured governments should be | allowed to form at the federal level and in each of the regions. Therefore, the former Soviet Union, Czechoslovakia, Yugoslavia, the United Arab Emirates or contemporary Pakistan cannot be conceived as federations proper. Tlxhd^federations do not normally allow for the unilateral secession of one or_^resubentrries. it a constitutionjvere to contain a secession clause, the secessiOTT^-Qnr region~ should besuYjectTd The consentjjfJjiejEerrtral g^errm^t^ndTmajority of ther^on^shtngtoTremainTn the federation." In factroTBy~theromtitution of the Soviet OlSoriTSt^'lEc^^ secession clause. Yet, the SU was not only a nonfederal entity, because it lacked the features of a democracy, but at the height of its powers the SU would have almost certainly used violence to stop such a unilateral secession from happening (Sunstein 1991). In some federations, the jurisprudence of the Constitutional or Supreme Court made clear that the states or regions cannot secede unilaterally. For instance, after the end of the Civil War, the US Supreme Court proclaimed that the US constitution 'looks to an indestructible union, composed of indestructible states' (Wheare 1963: 86). More recently, in a highly controversial ruling, the Canadian Supreme Court ruled that Quebec has a right to secede, but only when that secession is a negotiated one, that is, only after it has secured the agreement of the central government and the other provinces (Thomson 1999). Although lacking all the attributes of a federation, the Spanish Constitutional Court would most likely consider a unilateral move of the Basque Country to secede from Spain as a clear violation of Article 2 of the Spanish constitution. When political pressure builds up, a centre may well be tempted to negotiate on the terms of secession. However, on paper at least, the centre is too weak when a centre faces a set of regions which have the option to leave the federation unilaterally. If we apply the above criteria to the states of Western Europe we end up withjbreej?^^ "with most of these criteria. The classificatiori^fG^many "aha'SWifzerland as f^toanFBey^n^doubtTTBelgium is also a federation. True, the Belgian regions (regions and communities) are not directly involved in amending the federal constitution or in changing the rules that distribute competencies between the centre and the regions. Although the federal second chamber must give its consent to such changes, it is not an effective springboard of community and regional interests. However, constitutional amendments that alter the distribution of competencies between the centre and the regions must be taken with the consent of the two major language groups. Language groups strongly overlap with the two major Belgian communities (Flemish and French Community) and regions (Flemish and Walloon Regions). In this sense one can make a case that the major regions of the Belgian federation have a strong input in changing the structure of the Belgian federation. In contrast, Austria is not fully federal. For instance, jthe Austrian federal'second chamber is relatively weak as is the input of the regions in central decisions that affect their interests. Similarly, the political systems of Spain and the UK possess several attributes of a federation, but also lack some. For instance, Spain does not have strongly institutionalized mechanisms of 'intra-state federalism'. The Spanish Senate does not really function as a territorial chamber and there are few formalized interminister-ial conferences that produce binding decisions on the central and regional governments. In the UK, Scotland, Wales (and Northern Ireland) suffer from a lack of guaranteed input in amending the central parliamentary acts which determine their levels of autonomy. The British devolution settlements are not enshrined in a constitutional document, but in parliamentary laws ('Acts') which formally speaking can be amended or repealed at the sole discretion of the central parliament. In the absence of a supreme written vjuii^w.w^ii'u j i.MbiuujMl wmu ni.jlU/iMliJ/fl III V V LCI II LiLlI UJJC 10 constitution, there is no genuine constitutional court to watch over the compliance of Westminster and regional law with the central parliamentary j Acts specifying devolution. Later, I will argue that a more flexible, less for- | malistic interpretation of these arrangements warrants the treatment of f Spain and the UK within the family of federal or regionalized states in I Western Europe. Table 1.1 lists all nine characteristics of a federation, and applies them to six West European countries. CZ J Table 1.1 Standard characteristics of a federation applied to six West European States | —" Belgium Switzerland Spain Germany Austria UK / (1) Two orders V k V 4 Asymmetry f of government, (Sc, Wales^I) direct effect "") (2) Assignment of A V Asymmetry 1 'genuine no solid i 1 autonomy' constitutional in constitu- basis of tional way devolution \ (3) Intrastate Sufficiently Weak in Weak in Weak in ) federalism powerful powers powers powers and / (second but weak and weak weak / chamber) territorial connection territorial connection territorial connection (4) Supreme Supreme Weak V Weak No supreme written Constitution regional regional constitution, constitution but no explicit input in input in weak regional / only to be involvement amendment amend- input in / amended with of regions in ment amending consent of amendment * parliamentary / / / both orders of government acts (5) Umpire. V Supreme Court and the people Privy Council but no real constitutional court (6) Mechanisms V V Weak a/ V 1 for intergov- ernmental ■ i relations (7) Territoriality Dominant V V V • ■ V \ (Communities are partially based on personality principle) (8) Democracy • V V V (9)^ No unilateral 1 V V Note: V indicates presence of characteristic^) listed in the left column. 2. Confederalism, regionalism and regionalized states Scholars of comparative federalism have distinguished between various forms of governance that may embody some elements of federalism, without resulting in the formation of a federation (unions, constitutionally decentralized unions, federacies, associated states, condominiums, leagues, joint functional authorities; see Elazar 1987: 38-64; and Watts 1996: 8-9 for examples). A full discussion of all these terms is not needed here, but the distinction between a federation and a confederation, on the one hand, and a federation and a regionally devolved state (or a regionalized state), on the other, is useful. Compared with a federation, a confederation provides for a stronger position of the compounded entities. First, the entities that form a confederal arrangement retain their character as sovereign states. Second, in contrast with federal structures, confederations do not rule out the unilateral exit or secession of one or several of the federated entities. Third, confederal centres do not act directly upon the people, but prescribe the consent of the political institutions of the confederal entities first (for instance the member-state legislatures or their populations). Finally, decisions of the centre require the consent of all the states. Hence, the principle of unanimity fully applies (Croisat 1992). It is sometimes argued that the EU provides a good contemporary example of a confederal structure. The member-states of the EU act as sovereign states and are recognized as such in international organizations such as the United Nations, NATO or the OECD. The EU operates on the basis of a collection of treaties, and in principle member-states could decide to leave the EU. Treaty amendments require the consent of all the member-states by procedures of their choice (parliamentary consent, referendum, etc.). In highly sensitive policy areas, such as foreign policy or taxation, EU decision-making still requires the consent of all. However, in other aspects the EU has already surpassed the status of a confederation. Most common policy decisions are now taken by Qualified Majority Voting in the Council. The doctrines of supremacy and direct effect of EU law were established relatively early on in the process of European integration. While EU primary and secondary legislation (treaties, directives or framework laws) require further action by the member-state parliaments, this is not the case for EU decisions (a different type of EU law). The EU is not a federation, but it is already more than a confederation. Therefore, the best-known examples of confederal structures are historical: the Swiss confederation prior to 1789 and again between 1815 and 1845; the United Provinces ok-the Netherlands between 1579 and 1795; the German Bund between-1-815 and 1866; and the USA from 1781 until 1789, arguably even until 1865, as the southern states did not fully embrace US statehood until/the civil war had ended (see Forsyth 1981: 60-72). The long-term viability of confederal structures can be questioned. The historic examples demonstrate that confederations are likely to fall apart (Austria-Germany), to develop into federal states (USA, Switzerland and Germany which emerged from the North German Bund) or even into unitary decentralized states (the Netherlands). Confederal centres are weaker in relation to the member-states of a confederation than federal centres in relation to the regions of a federation. However, federal centres are weaker in relation to the regions of a federation than the centres of a regionalized state in relation to their regions. I will refer to regionalized states as states that exemplify some form of 'regional devolution' (Keating 1998a: 113). Regional devolution is the result of a process of 'regionalism', a term that is couched in even more ambiguous terms than federalism. Regionalism refers to a process which leads territorial subunits within or across existing sovereign states to increase their influence. That process may have a socioeconomic, political or cultural driving force or may be a combination of all these factors. As such regionalism is in part a bottom-up process, but the consent of the centre is needed to increase the levels of regional autonomy. Ultimately, the centre may be willing to recognize a greater role for the | regions if that can safeguard or expand its political support across the state. Strong forms of regionalism may produce a regionalized state (or transform the latter into a full-fledged federation). Regionalized states have strongly developed local or regional tiers of government with directly elected councils, j Unlike the regions of a federation, the regions in a regionalized state remain j subordinate to the central government. The centre can increase, decrease or I even suspend or withdraw the regional levels of autonomy without requir- j ing the consent of the regions. In addition the scope of the devolved powers \ (but in this regard Scotland is an exception) is not as extensive as is the case j i in a federation. Spain is such a regionalized state, and in most of the charac- ■ teristics that were listed above already approximates to a full-fledged federa- ' tion. In Spain, the central government has negotiated Statutes of Autonomy with each of the regions (Autonomous Communities). These statutes cannot be unilaterally repealed, although they are bound by constitutional principles. Changing the constitution as such does not require explicit regional consent. The UK is yet another example of such a regionalized state. ; Successful regionalization efforts thus far only affect about 15 per cent of the j population, that is, UK citizens who live in Scotland, Wales or Northern i Ireland. In principle, the British Parliament could expand or reduce the ] autonomy of Scotland, Wales or Northern Ireland without their explicit 'j consent (In reality, it is assumed that at least Scotland and Wales would not lose all of their recently gained autonomy without the consent of the people j who live in both regions).. Regions of a regionalized state stand in a weaker position relative to J the centre than regions in a federal state. However, they are in a stronger position than the subnational entities in a unitary decentralized state. Like regionalized states, the subnational entitles owe their strength to the centre. Most likely these subnational entities also have directly elected councils and executive bodies. However, unlike the regions in a regionalized state, they have fewer legislative, administrative and/or fiscal powers. Compared with regionalized states, the 'making' of regional boundaries is also more likely to be the result of top-down regional planning than of bottom-up popular demand. Admittedly, there is no real cut-off point between both groups and, in Western Europe alone, the number of unitary decentralized states is still quite broad, comprising three different groups of countries. At least two of them are moving in the direction of a regionalized state, therefore I shall take them as the starting point of this overview. The first group consists of Italy and France. At present, the Italian regions are about to achieve the same level of regional autonomy as the Spanish autonomous regions in the 1980s. Shortly after the Second World War, 5 'special' regions were recognized with specific cultural or geographic features. The regions of Trentino-South Tyrol, Aoste and Friuli-Venezia-Giulia contain respectively German-, French- and Slovenian-speaking minorities. They received a 'statute of autonomy' which conferred some fiscal, socioeconomic and most importantly cultural autonomy upon them. The geographically isolated island regions of Sicily and Sardinia received some form of autonomy as well, focused more on socioeconomic matters (Palermo 2005: 184-5). A further 15 'ordinary' regions did not come into operation until the 1970s. Their autonomy did not increase significantly until the late 1990s. None of these regions has clear historical roots. Their borders were in fact 'created' in the 1930s for statistical purposes. The collapse of communism in 1989 sent Shockwaves through the Italian party system. In some respects it also reawakened the so-called Northern Question. GDP per head of the population was almost twice as high in the Northwest and Northeast of Italy as in the South (Gold 2003: 66). The Lega Nord tapped into the feelings of discontent that emerged from these large socioeconomic divergences. It emerged as an important political force in the North. It called for more Northern autonomy (provocatively even for the secession of Padania, a fictitious state which encompasses the regions to the north of the Po river). In-this way, the North would be able to keep a larger share of national wealth for itself and to stop financing 'pork barrel' projects which exclusively benefited the South. Federalism or 'devolution' has been on the agenda since the mid-1990s. At this stage, all regions have been offered a statute of autonomy (but in July 2004, only two ordinary regions had effectively enacted one). The regions have directly elected presidents and assemblies. Some forums of intergovernmental relations have been set up to incorporate the leading regional figures into central decisions which affect their interests (such as the allocation of EU Structural Funds). However, most of these mechanisms are ad hoc and generate non-binding decisions. The Italian Senate does not operate as a territorial chamber. Furthermore, the levels of regional autonomy remain relatively limited, both in policy and in fiscal terms. It remains unclear in which policy domains the regions have the final say. The Italian Constitutional Court has played a significant role in clarifying that matter (Palermo 2005: 188). Since the so-called Bassanini laws have been implemented (1999), central parliament cannot simply overturn regional laws. A referendum in 2001 put Italy on course for further devolution, possibly even federalism (Amoretti 2002, 2004). However, disagreements on the nature of devolution (should all regions receive the same set of powers) and widespread resistance to devolution among one of the central government's coalition partners (the right-wing but also 'pro-centralist' Alleanze Nationale) slowed down or even reversed the process of devolution. When this book went to press, the Italian government was introducing legislation that would 'undo' some of the reforms which the Italian population had agreed to in 2001. In this sense, it makes sense not (yet) to include Italy in this comparative study, that is, until more clarification is given on the distribution of competencies, the autonomy of the regions and their participation in intergovernmental channels. The possibility for each region to negotiate a statute of autonomy with the centre is in fact very much inspired by the Spanish example. At least in the medium long term this may give rise to 'a highly differentiated, asymmetrical regional system' (Palermo 2005: 193). Next to Italy, the French regional levels of government have strengthened considerably in France since Francois Mitterrand enacted his decentralization reforms in 1982. In 1991 regional autonomy was increased in Corsica and has continued to expand there in the wake of small-scale nationalist violence (Loughlin and Seiler 2001: 196-7). Yet, the political autonomy of the 22 French regions is constrained in several respects. For instance, a centrally appointed prefect (governor) coordinates central administrative powers and supervises the regionally elected bodies in the exercise of their devolved powers (Keating 1998a: 113). In terms of the scope of their powers, regions remain the 'weaker cousins' of the 96 departments and more than 36,000 local governments. Regions only have some responsibility in economic development, transport, secondary education, tourism, training and culture. Most regions employ less than one hundred full-time salaried staff. Consequently, they rely on the assistance of employees at the departmental or local levels. Their budgets (which only consist of grants) are dwarfed by the budgets of the departmental councils, despite the fact that they cater for larger population groups. For instance, in 1995, the budget of the Nord department was twice the budget of the Nord Pas de Calais region. This is so despite the fact that the region catered for nearly four million inhabitants, and the department only for two and a half million (Cole 2005: 122-3). Departments have a history that goes back to 1790. Their political structure is supervised by centrally appointed prefects. Departments are governed by councils with an (indirectly) elected basis. Departments exemplify a form of 'functional decentralization', seeking to uphold France's Jacobin administrative philosophy of 'a single and indivisible Republic'. If we leave aside Corsica and Brittany, all of the French regions have strong artificial roots; they are central bureaucratic creatures. In 2003-2004, a constitutional reform finally recognized the regions as one of four levels of local authority. The same reform also enabled the regions and departments to bid for additional functions. The central parliament would be responsible for decentralizing these functions (most likely transport infrastructure, such as airports). After a period of five years it would assess whether their decentralization should be made permanent. However, the constitution does not tolerate wide-scale asymmetry (if a function is decentralized, it ought to be decentralized to all similar types of subnational government). In July 2003, the voters of Corsica were offered the possibility of merging their departmental structures (Haute Corse and Corse-Sud) into a single regional unit. The Corisican voters narrowly rejected it (Cole 2005: 129). The second group of unitary decentralized states is made up of the Scandinavian countries (but it excludes the special status of Greenland in relation to Denmark). The Scandinavian countries have a tradition of unusually strong local government. If we set aside the Swiss municipalities, Scandinavian local governments are stronger in administrative and fiscal capacity than their counterparts in all the other West European countries. Therefore, many analysts of fiscal federalism lumped the Scandinavian countries together with the federal states, because their local governments absorb such high shares of total public expenditure (Ter-Minassian 1997; Fossati and Panella 1999; Wellisch 2000). However, it would be wrong to consider these states as federal for three reasons. First, a considerable amount of local government tasks relate to the administrative implementation of central government welfare legislation. Municipalities have a limited input in shaping that legislation. Second, the activities of local government outputs are more often subject to central oversight. Third, the size and number of municipalities can be adjusted with relative ease for reasons of cost-efficiency or urban planning. For instance, the number of Swedish municipalities was reduced from approximately 2,500 in the 1950s to 289 in 2001. Although this may have improved the long-term viability of the localities, such a reform also illustrates the difference between the constitutional position of a region in a federal state and that of a municipality in a decentralized unitary state (Lindström 2001: 319-42). Changing the borders of a region'in a federal state, or merging existing regions cannot normally be accomplished without some sort of regional consent. Creating the Swiss Jura canton or merging the German Länder of Berlin and Brandenburg required the consent of the affected regions (only with success in the first case). The final group of countries is the easiest to classify. It consists of countries such as Greece, Luxembourg or Portugal (excluding Madeira and the Azores) Pakelta socldirich stodit S0200 BRNO CID with respectively weak or no regional tiers of government at all (Loughlin 2001). 3. Multilevel governance In the previous section, I identified three major groups of states on the basis of the institutional strength of the regions therein: federal states, regionalized states and unitary decentralized states. Some may dispute the criteria that were used to delineate these categories, but at least readers should know what distinguishes them. The absence of clear, universally agreed cut-off points, in particular between regionalized and unitary decentralized states, is due to the frequent understanding of regionalism as a continuum and not as a taxonomy. In this regard the more recently developed term of multilevel governance (MLG) can be used as an umbrella concept in which all forms of decentralization find their place. Two of the authors who popularized the term, Liesbet Hooghe and Gary Marks, defined MLG as the 'dispersion of authoritative decision-making across multiple territorial levels' (Hooghe and Marks 2001: xi). Initially, the concept of MLG was developed to provide a better understanding of the EU, a political system which until the 1990s was primarily interpreted through the contrasting lenses of intergovernmentalism and (neo-)functionalism (Rosamond 2000). MLG provided a middle way between both theories. It acknowledges the central position of national governments in European integration, but also argues that 'authority and policy-making are shared across multiple levels of government - subnational, national and supranational' (Hooghe and Marks 2001: 2). Collective European decision-making has weakened individual state autonomy, because some decisions no longer prescribe unanimity voting but nonetheless enforce common rules across the EU. Furthermore, MLG theorists make a strong claim that subnational interests are not necessarily mediated through national governments, but may find direct access to the European policy arena. Based on their analysis in Multi-Level Governance and European Integration, Liesbet Hooghe and Gary Marks seem to distinguish MLG from a federal stateformat. They point at the lack of a formalized supreme constitutional framework which specifies the legal ends (central competencies) of integration, and the absence of EU statehood warranting the legitimate use of violence. Echoing the confederal traits of the EU, they argue that 'national governments in the member-states have greater powers of self-determination than constituent units in any existing federal state, and while the territorial units within a federal regime tend to have similar subnational political systems, the domestic political systems of the member states vary greatly' (Hooghe and Marks 2001: 37). However, in their more recent work, Hooghe and Marks have broadened the concept of MLG to include any type of political system that provides for multiple territorial tiers of government (Hooghe and Marks 2003). Federations then simply become a specific species of MLG, a different brand of MLG from the one that characterizes the governance of the EU. As they see it, federations form part of a type of MLG (Type I MLG') that is characterized by four major characteristics: (1) a limited number of territorial jurisdictions (central, intermediate or regional and local); (2) the long-term stability of these jurisdictions, as territorial boundaries cannot be easily adjusted in light of emerging functional desires; (3) the replication of comparable institutional structures at each level, reflecting some form of balance between an executive, legislature and judiciary; and (4) the non-intersection of territorial memberships, that is, the smaller units are neatly contained within the borders of the larger ones. The authors contend that these characteristics are not exclusively confined to federations, but could be used to understand policy-making in regionalized or decentralized unitary states as well. The distinguishing element between this and a second type of MLG ('Type II MLG') relates to whether public policy coordination takes place within traditional, hierarchical territorial structures, or rather in flexible, innovative networks with an ad hoc and functional character. Such networks are more frequently based on voluntary individual membership, although they could also bring together municipalities or regions which share a concern to respond to specific needs, such as a more efficient provision of public utilities like gas or electricity (Hooghe and Marks 2003). The division of MLG along two types broadens our understanding of public policy-making beyond the traditional framework of the state. However, there are two difficulties in relying on MLG instead of a more 'traditional' federal-regionalized-unitary decentralized framework. First, as federations are fully contained within MLG Type I, there is insufficient clarification as to how different degrees of regional institutional strength should be operationalized (but see the useful Appendix 2 in Hooghe and Marks 2001:191-209 for some guidelines on how this could be done). Second, and arguably of greater importance, federal-type structures can also display many of the attributes of the so-called MLG Type II forms of governance. A federal centre or the regions of a federation can stimulate the creation of more functionally oriented associations. For instance, in Switzerland each of the 26 cantons (regions) has substantial policy and fiscal autonomy. Yet, the smallest cantons in particular have been forced to cooperate so that certain utilities can be provided in a more cost-effective way without having to rely on the support of the federal government. This has led to the formation of functional, ad hoc cooperative networks with a more informal and fluid character. These networks exist in parallel to the more widespread intergovernmental structures that we find in most federal states. Frey and Eichenberger introduced the concept of Functional Overlapping Jurisdictions (FOCJ) to denote such functional cooperation areas that cut across traditional territorial boundaries of the state (Frey and Eichenberger 2001). ^urnyuiuuve reueransm ana Kegionansm in western turope Zl The spread of MLG Type II may vary from one federation to another, but also fluctuates from one region within a federal state to another. We are likely to find more of MLG Type II in federations that kept a 'competitive federal design' than in federations that have sought to establish harmonized policy standards across the federation. In the former, the central level can be rolled back more easily, or, as I will discuss in the chapter on pubhc'pbTicy-making, the regional governments may have had more veto powers in the centre to prevent the centre from playing a more active role in providing public goods. In sum, even when seeking to analyse 'MLG Type II', federalism becomes an important independent variable, the varieties of which can help us to broaden our understanding of public policy-making in federal and non-federal states alike. 4. Conclusion In this chapter I sought to clarify some key concepts in comparative federal studies. In essence, federalism is a normative concept which entails a combination of self- and shared rule. In this sense, federalism is of little analytical value. When I use the concept of a 'federal' state it is-meant_to refkctjrpolitical structure which does not only embody federalism as a philosophy, but also complies with nine specific institutional features. Perceived accordingly, a federal state is always a federation. Next to denning federalism and federation, I identified the difference between a confederation, federation, regionalized and unitary decentralized state. As stated in that order, these political systems range from possessing the weakest to the strongest political centres. In a confederation, the component entities retain full sovereignty as states and the centre is so weak that it cannot act without the consent of all the states. In regionalized and unitary decentralized states, the regions are in a much weaker position They owe their autonomy to the centre. The centre can repeal that autonomy and put itself m a position of hierarchical control whenever it enters a conflict of interest with one or several of the regions. A vague consensus exists on how to define a federation, but the cut-off point between a regionalized and a unitary decentralized state remains unclear. This is so because the level of decentralization for both groups of a SI*, Tf ti0ned °n 3 COntinuum- Their *gions are weaker than in a federation, but more so in a unitary decentralized than in a regionalized hnnnl\TTfd St3te' regi°naI §ovemment merges as the result of bottom-up calls for more autonomy. Such demands may have their roots in socioeconomic developments (such as the call for more regional autonomy for the North of Italy). Alternatively, a call for regionalism Say be fueled by minority nationalism. Devolution for Wales and Scotland, and also for the Spanish minority nations (the Basque Country, Galicia and Catalonia) are a good case in point. In Spain, at least, the other regions, which lacked strong historic roots, quickly demanded similar levels of autonomy. In this sense, regionalism spilled over from the minority nations to the non-historic regions. By comparison, in a unitary decentralized state, regions are created more often as the result of top-down planning. This is the case for most of the French and Italian regions or the Scandinavian municipalities. In some cases, regionalism and 'top-down' decentralization may be combined within the framework of a single state. For instance, Scotland, Wales (and Northern Ireland) have turned the UK into a regionalized state. However, the proposed (and so far failed) 'regionalization' of England is the result of top-down planning. Similarly, with the exception of Brittany and •Corsica, the French regions have weak historical roots and lack sufficient powers to fight against the more strongly embedded departments. Unlike in Britain, Brittany has not more autonomy than the other regions. In Italy, South Tyrol exemplifies a form of minority nationalism, and the special regions have specific historical or cultural reasons that qualify for regional autonomy. Yet, resistance to a form of asymmetric devolution is high, particularly among the Southern regions which need the support of the North for their economic survival. The regions of a federation possess genuine policy autonomy that is constitutionally embedded. A federal constitution cannot be amended unilaterally by the centre. Although the regions in a regionalized or unitary decentralized state do not possess that degree of autonomy, their level of autonomy exceeds that of a region in a unitary decentralized state. More often the regions in a unitary decentralized state experience competition from other subnational tiers of government such as provinces, departments or even local governments. For instance, in France, the regions are weaker in fiscal, legislative and administrative capacity than the departments. The departments often serve to extend central political power into the periphery. In Italy too, regions experience some competition from the provinces and the local governments, which they do not yet control. The overview above clarifies why the inclusion of Austria, Belgium, Germany Spain and Switzerland as federal or regionalized states into this comparative study makes perfect sense. The regionalization of the UK is only ' confined to a small part of its territory.1 However, the scope of the legislative and executive autonomy that was granted to Scotland (less so to Wales) is so extensive that the UK deserves to be included in the analysis. Although discussions are ongoing on extending the legislative, fiscal and administrative powers of the Italian regions, the level of decentralization has reached that of the Spanish regions in the 1980s. Therefore, I have not (yet)included Italy in the comparative analysis.