WILLIAM RIKER ON FEDERALISM: SOMETIMES WRONG BUT MORE

RIGHT THAN ANYONE ELSE

David McKay, University of Essex

Paper before the William H. Riker Conference on Constitutions, Voting and

Democracy, sponsored by the CNISS and the Center for Political

Economy, Washington University, St Louis, December 7-8^th, 2001

William Riker published on federalism over a long period starting with his 1955 APSR article 'The Senate and American Federalism,' and culminating in the 1987 volume, The Development of American Federalism. Between these dates he produced the seminal 1964 volume Federalism: Origin, Operation, Significance, an updated and condensed version of which appeared in the 1975 Handbook of Political Science (Riker, 1955, 1964, 1975, 1987). In addition he produced a small number of technical articles with graduate student collaborators (Riker and Schaps, 1957; Riker and Lemco, 1987). Two features of this opus stand out. There is first the project of establishing a general theory of federalism organized around an attempt both to explain the origins and the sustainability of federation as a form of government. Second, Riker repeatedly returns to American federalism and in particular how political institutions have mediated federal/state conflicts in American history. It should be added that for much of the latter part of Riker's intellectual career, his interests were diverted to other concerns, and notably what he called heresthetics or the ways in which rationally motivated political actors could manipulate institutional rules to their advantage. But here too, issues of federalism loom large, as many of his examples are drawn from American history and the center-periphery conflicts that dominated much of the first eighty years of the Republic (Riker 1982; 1987).

This paper will concentrate on Riker's attempt to build a general theory of federalism. Part One will provide a critique of his explanation of the origins of federal unions and show how his original claims need to be qualified and refined in the light of recent work on the genesis of the European union and of the Australian and Indian federations. Part Two will address his (incomplete) explanations of why, once the conditions that account for the original federal bargain have passed, many federal systems persist. In the American and Canadian cases federal cases, for example, the external threats that Riker invokes as the original federal raison d’être have been absent for well over 100 years. In other cases, absent the threats, and the federations fail (Yugoslavia, Czechoslovakia). Constitutional arrangements and in particular the ways in which they broker preferences and produce indeterminate outcomes are central to explaining the subtle gradations of federal systems whether these be defined in terms of viability or of the degree to which they are centralized. Part Two will examine three institutional dynamics that help shape federal systems: the operation of upper houses, the role of institutions in restraining the popular will and the structure and functioning of party systems. By necessity, this discussion will be brief, and the analysis often schematic. Throughout, however, the aim will be to use empirical examples from federal systems that were not the main focus of Riker's work not only to show the strength of his original insights, and also to show how in many instances they need to be qualified and refined.

Part One

On the Origins of Federal Unions

In Federalism: Origins, Operation, Significance, Riker was characteristically blunt in his dismissal of the existing literature on federalism which he considered ideological rather than scientific (Riker 1964, Chapter 2). He was right. Almost all of the theoretical writings on the subject up to that time had been infused with assertion, unsubstantiated claims and value judgements (see, in particular Friederich, 1968, Deutsch, 1957)[1]. Riker is particularly scathing about those who explain the emergence of federations in terms of a search for economic growth or for democracy and individual freedom. The most casual of empirical observations quickly refutes such claims. Impoverished Mexico is compared with prosperous United States; freedom in Canada is contrasted with oppression in Yugoslavia and the Soviet Union. Within political systems, federal arrangements may appear to serve the cause of freedom and economic development (modern Germany) or may appear to inhibit them (Mexico, the American South down to the 1960s). Even greater scorn is cast on what he calls the ideologists or those who explain federalism in terms the act of 'federalizing.' As he puts it:

A tremendous amount of propagandizing and even political organizing has been based on the mistaken premise that somehow, if people just work hard enough for it, federation will occur. It is perhaps unkind to disturb such naïve faith, but the hope of scientific enterprise is that the more people know, the more effectively they can act. A particularly unfortunate example of this naivete is to be found in Friedrich (1968), where it said that federalism is a process of federalizing, as if such a thing comes about by some kind of magic without rational human calculation. (Riker 1975, 131).

More recently, Malcolm has made the point just as starkly by noting that federalist ideologues appear to believe in a 'kind of cartographic mysticism that intuits that large areas of the map are crying out to merge as single geopolitical units' (Malcolm, 1995, 53).

Riker's alternative is, of course, to characterize federalism as a rational bargain between prospective national leaders and officials of constituent governments who come together for the purpose of creating a larger territory so as better to facilitate the levying of taxes and the raising of armies (Riker 1964, Chapter 1). The two conditions that Riker claims must always be present before a successful bargain is struck were, first, a desire on the part of the of those offering the bargain peacefully to expand territory by combining constituent governments into a new political entity in order to meet an external military threat or threat to internal order.[2] Second, for those accepting the bargain some sacrifice of political control is exchanged for the promise of security provided by the new federal government. Both in his 1964 volume and the subsequent 1975 version, Riker is certain of the empirical validity of his theory. As he claims:

'In order to prove this hypothesis I have examined all instances of the creation of federalism since 1786….For those federalisms that have survived, I am able to show that the two conditions existed at the origin; and for those that failed, I am able to show that either the conditions never existed, or they existed only momentarily. Though such evidence does not constitute absolute proof of the hypothesis, it comes as close to a proof as a non-experimental science can offer' (Riker 1964, 10).

From the perspective of 2001, two features of this not immodest claim stand out. First, while it has been widely critiqued as reductionist, none of the critiques is substantively sustainable (see, for example, Dikshit, 1971; 1975). Second, and perhaps more telling, no one has come up with a theory of federalism that is remotely as ambitious or as powerful. Indeed Riker's theory remains, almost forty years later, the only theoretical perspective on the subject worthy of the name.[3] This said, it is all too easy to demonstrate the sometimes cavalier way in which Riker uses empirical material. As a result, Riker's framework needs to be qualified and amended so as to retain validity for all cases. What follows is a brief analysis of the early critiques followed by a discussion of three hard cases (Australia, the European Union and India) that show the ways in which the Riker thesis needs to be modified and updated.

Critiquing Riker

Potentially, the most telling critique of the Riker Schema on the origins of federations concerns his insistence that the threats to external or internal security must always be present at the inception and must always be military/diplomatic in nature. There are three dimensions to this critique - the universality of threats, the level of threats and the nature of threats. Claims that the theory is flawed because all countries actually or potentially face threats so this condition must be present in all cases (see Dikshit, 1975, 222-227; King, 1982), can easily be dismissed. External threats may result in a number of outcomes including annexation, treaty or war (Mearsheimer, 2001). Only when certain special conditions are met will threats result in federalism. These must include the presence of the conditions of the Rikerian bargain, i.e.: a perception on the part of centralists and localists that entering into a mutually beneficial peaceful and negotiated exchange of power resources is necessary to avert or reduce the threat. Whether or not this occurs will depend on the particular circumstances of particular cases. Clearly the level of threat will be part of this calculus and will help determine the particular arrangements arrived at. Threat of annihilation or conquest can throw up the most unlikely of federalisms, as was almost the case in 1940 when Churchill and the entire British cabinet proposed an 'indissoluble union' between Britain and France (for a discussion, see Jennings, 1940, MacKay, 1940).

Discussion of the level of threat can also be conflated with the nature of the threat. The proposed Anglo/French union was designed to cement the two countries together in their efforts to defeat Nazi Germany. Should this have been accomplished the incentive to federalize further and deeper would almost certainly have been low and the union would have dissolved. In contrast, the Swiss case shows how a small country surrounded by hostile powers for a period of many centuries found it necessary to maintain and strengthen its federalism by creating central institutions capable of sustaining a permanent national army. In time these institutions came to perform other functions such as regulation of internal and external trade, social security and transportation. We will return to the relationship between institutional design and the allocation of central and provincial powers in the next section.

When discussing why Riker insisted that the threats had to be military or diplomatic in nature, two points stand out. First, Riker was reluctant to include economic improvement or the protection of cultural values as necessary conditions for federalism because he maintained that such claims were easy to demonstrate as empirically false. Only federalist ideologists made such assertions. But Riker's rejection of the economic and cultural is too restrictive. If instead of economic improvement or cultural solidarity the motives of the federalizers are expressed in terms of a need to avert an economic or cultural threat that is perceived to challenge the interests of dominant political coalitions, then the economic and cultural can placed in the same analytical category as the military and diplomatic. Often some combination of two or perhaps all three of these threats will be present. Such a modification has two distinct advantages. First, it permits the inclusion of some federal-like states that were rejected by the original thesis because no meaningful external threat was present. The European Union is such a case. Second, it makes more convincing some federal designs that Riker originally insisted were premised on military threat, when it is patently the case that economic and cultural factors were as important. Australia fits this category.

Another challenge to the Riker schema would be the existence of a functioning federation - empires and autocracies apart - where it is difficult to identify an originating federal bargain based on negotiation between centralizers and territorially defined provincial interests. In the case of post partition India, for example, centralizing Indian elites imposed federalism with little reference to provincial interests. Indeed, the boundaries of the new states had, in many instances, little or no historical, cultural, linguistic or ethnic basis. They were, instead, largely the construct of the framers. Let us look at these cases in more detail, albeit in a highly summarized form.

Three Difficult Cases

1. The European Union. Writing in the 1960s and 1970s Riker was insistent that the EC did not qualify as a federation. No external or internal threat was discernible. As he put it: 'For… [a European Union] to appear there must be some significant threat. And in the absence of a threat large enough to render the federal bargain mutually profitable to the participating governments, there is nothing that will bring… [such a union] about, no matter how much people wish for…[it] to happen' (Riker 1975, 130-131). He remained sceptical even during the 1990s - although he did eventually concede that the advantages associated with the removal of trade barriers might be sufficient to provide incentive enough adopt federal like arrangements in Europe (Riker, 1996).[4] Yet there is now broad agreement that with monetary union, the EU can be classified as a species of federal state (see the discussion in McKay 2001, Chapter 2). And while at the time of Maastricht direct military/diplomatic threats were notably absent, economic threats were palpable, some of which had important implications for the prevailing political order in a number of member states. Among these the most important was the perception among bankers and finance officials that some external discipline was necessary to control inflation, and that only EMU would guarantee this. Such thinking dominated discourse in Italy, Belgium, Greece, Portugal, and Spain (Dyson, 1996; Dyson and Featherstone, 2000). In the latter three states, EU membership in general and EMU in particular had the added advantage of cementing fragile democracies into a European framework of law and commerce built on principles of individual freedom and democratic representation. French thinking too was influenced by the dislocating effects of recent economic history including the inflationary effect of flirting with economic nationalism during the early 1980s and the wobbles experienced by the franc fort policy in 1987. Most of the voluminous literature on this subject dwells on the motives behind German participation in EMU or on the dynamics of Franco-German bargaining at Maastricht and after (see in particular Dyson, and Featherstone, 2000; Moravscik, 1998). Why, most observers ask, did the Germans sacrifice their precious Deutschmark and run the risk of importing Italian style inflation? In retrospect there is a relatively easy answer: European Union would, for the first time in 40 years, allow Germany to play a diplomatic (and eventually military?) role commensurate with its economic status. And by insisting on restrictive convergence criteria, a central bank free from political control and a Stability and Growth Pact designed to punish inflationary recidivism, the Germans would create a system modeled on German monetary discipline rather than Italian financial delinquency. If they could achieve diplomatic aggrandizement while forcing the rest of the EU to adopt German monetary discipline they would be in a win win situation. The evidence in support of this interpretation is impressive (Not least the role played by German bankers and finance officials at Maastricht, Dyson, 1996, Chapter 5). This interpretation can be expressed nicely in Rikerian terms. All the participants at Maastricht bar the British and the Danes were willing to cede elements of national control to a federal authority because the benefits of doing so exceeded the costs. Not all the motives were the same, but all fell the right way. In some cases, the calculations may have been objectively correct and in other cases objectively wrong. Italian inflation has all but vanished in the last half dozen years. German membership of EMU may have borne much greater economic costs in the form of deflation than was calculated during the early 1990s.

2. Australia. Riker made two bold claims: that Australian federation was precipitated by German, French and Japanese imperialism in East Asia and the Pacific, and that because these threats were relatively weak, it was necessary only to create a relatively peripheralized federation that reflected the strength of provincial as opposed to national loyalties. He also conceded that economic factors were, if not dominant, then certainly influential. More recent scholarship shows, however, that the Australian federal bargain resulted from a complex mix of military, economic and cultural factors. Militarily, the greatest threat to Australia occurred eighteen years before the creation of a federal state in 1901, when in 1883 both France and Germany threatened to annex the New Hebrides and New Guinea. In order to forestall annexation the Queensland Premier declared the eastern part of New Guinea British Crown property. Although supported by the other Australian states, the British refused to support him (Dikshit, 1975, 101).[5] Later events, including Japanese expansionism in the Pacific and East Asia, were never to assume the same proportions as this episode. Culturally, the Australians felt threatened not by a Japanese or Chinese invasion both of which were extremely remote possibilities, but by a perception that the essentially British nature of the colonies would be ‘corrupted’ by Japanese and (especially) Chinese immigration. They were also fearful of immigration from New Guinea and the pacific islands. (Denoon and Wyndham, 2001, 564-569). And although by the 1880s state parliaments had enacted restrictive immigration laws, strong support for Australian wide laws to protect a ‘White Australia’ policy existed. Only a federal authority could provide such a law and prevent backsliding by state such as South and Western Australia whose economies needed cheap mine labour.[6] Most historians agree that the precipitating factor in the genesis of Australian federalism was the externally induced economic depression of the 1890s, which converted free trade bankers and financiers into economic nationalists. With the collapse of the Argentinean speculative bubble, London lending institutions also withdrew credit from almost as frenetic investment bubbles in the disparate Australian states. The resulting economic dislocation was sufficient to persuade both financial and labor interests that the sub continent needed to rid itself of inter-colonial trade barriers and retain the option of protecting all of Australia’s trade against outside competition (Dikshit, 1975, 102-103, Quick and Garran, 1901). The resulting constitutional settlement although described by Riker as peripheralized, transferred to the federal government all customs and excise taxes. Other taxes were to be shared by the states and the federal authorities concurrently. As customs and excise taxes were the main revenue sources for the states this transfer amounted to a quite dramatic centralization of power in the new federal government. In addition the federal authorities were effectively given control of both immigration and defense.[7] It is clear that Australian federalism was not simply a reaction to an external military threat. It was a reaction to a combination of external military, cultural and economic threats. The resulting constitutional settlement reflected these fears.

3. India. Riker explains post-partition Indian federalism in terms of a need to offset external threats from Pakistan and internal threats from the princely states that historically had controlled large tracts of territory before and during the British colonial period (Riker 1964). While security concerns loomed large in the creation of the Indian state, it is not obvious that they were part of a Rikerian bargain between central and provincial interests that required specifically federal arrangements. Indeed, the states were not even consulted on the shape of the new constitution. Instead it was the product of the Constituent Assembly on which the provinces had no representation. (Washbrook, 2001, Moore, 2001). Moreover, the Assembly then proceeded to create states some of whose boundaries had no relationship to any previous historical entity. For example the state of Bombay was entirely a new construct stretching as it did from the Pakistani border all the way down to Kerala. As if to emphasis the arbitrary nature of this project the constitution permits the federal government to alter states boundaries without their permission. They were radically re-drawn in 1956 and boundary tinkering has continued unabated ever since (Mitra, 2000, 44). This is not to say that the boundaries are completely arbitrary: they do in most cases approximate to ancient ethnic, geographic and linguistic regions. But the geographic fluidity of Indian federalism both at its inception and ever since casts into serious doubt that it was created as part of a central/local bargain. Rather than the 600 princely states ceding power to the center in exchange for their peaceful participation in the new system, within 18 months the princes had either acceded in their own dissolution or had withdrawn into private life in exchange for cash. In the solitary case of the Nizam of Hyderabad who resisted, the state was forcibly absorbed following military intervention. With the redrawing of state boundaries in 1956 all trace of the old Mogul states passed into history (Dikshit, 1975, 117-118). Other egregiously centrist features of the Indian constitution are worthy of note. State governors are not only appointed by the president of India on the advice of the prime minister, they can also submit controversial state bills for the presidents’ assent. Most importantly, governors, who are effectively the agents of the governing political party in Delhi, can request that state governments be suspended and replaced with direct rule from the center. Following the most serious challenge to central power in 1975-77 a national emergency was declared that effectively replaced federalism with unitary rule (Mitra 2000, 44-45). The suspension of state governments has periodically been sanctioned in Kashmir, the Punjab and in the isolated Northeast (Manor 1998). None of this is to deny that in modern India a form of federalism prevails or that central/local bargaining has decided the precise form of that federalism (Manor, 1998, 22-24). How that bargain was concluded after 1977 has yet to be researched using a choice framework. It is simply to claim that the original federal design owed little to a bargain in the classic Rikerian sense. Instead, Indian constitutional arrangements were quasi-unitary in nature and very much the product of a centralizing Congress Party determined to retain sufficient power in Delhi to ensure the territorial survival of the new country. Put another way, the resulting constitution failed to meet Riker’s minimal condition for inclusion in the federal category – that ‘both kinds of governments rule over the same territory and people and each kind has the authority to make some decisions independently of the other’ (Riker 1964, 5). This condition cannot be met if the central government can unilaterally change state boundaries and suspend state constitutions by presidential (prime ministerial) decree.

What these cases demonstrate is that in his determination to advance an inclusive or scientific explanation of the origins of federations, Riker oversimplified. His insistence that security had to be defined exclusively in terms of military or diplomatic threat was almost certainly in response to those who characterized federalism in terms of its association with economic or cultural (usually labeled ‘common interest’) desiderata. While it is patently obvious that federalism does not lead to such outcomes (or to the advance of freedom), there is no logical reason why averting economic or cultural threats should not be part of framers’ bargaining calculus, if these threats were seen by central and provincial politicians to upset the prevailing political order by challenging the interests of dominant elites. This refinement of Riker’s work differs from earlier critiques that tend to infer from the absence of military threat, the importance of other social and economic ‘factors’ that may or may not represent a threat to ruling coalitions (Dikshit, 1975, Lemco, 1991, 15).

Riker was also over eager to include in his federal taxonomy all those states that were labeled federal and had come into being in the context of external threat, even when, as in the case of India they failed to meet the conditions of his own definition of federalism. But, crucially, his simple axioms on the origins of federal states remain the most complete – and the most influential – to date. As such they continue to inform empirical research almost 40 years after its first formulation (Filippov, Ordeshook and Shvetsova, 2001, McKay, 1999).

Part Two

On the Sustainability of Federations

Riker’s explanation of why some federations survive while others fail is much less complete than his thesis on the origins of federalism. Indeed he does not attempt to provide such a theory and instead supplies insights drawn mainly from his familiarity with the American and Canadian cases. In the 1960s he saw federalism as a mild nuisance mainly because it had helped preserve the racist practices of the Southern states (Riker, 1964, Chapter 8). By the 1980s, he had shifted ground and became infatuated with the Madisonian notion that federalism was one of the conditions of the Compound Republic, or provided constitutional constraints on populistic politics. In particular he contrasted what he considered the unfettered constitutional authority of British governments with the constitutional constraints of the American system that made difficult the business of building coalitions that result in zero sum situations (Riker, 1982, Chapter 10). By implication, at least, federalism (along with the separation of powers and other veto devices) is more likely to check populist policies and the less centralized the federation the stronger this guarantee will be. By implication, the ‘ideal’ constitutional design is that which over the long term accurately reflects the balance of central and provincial interests, while at the same time providing a bulwark against the adoption of ‘populist ‘ politics that will offend territorially defined minorities. According to Riker, the American constitutional system has, the civil war apart, roughly achieved this balance (Riker, 1982, Chapter 10).

Although Riker shows how this has changed in the US over time he fails to develop the argument into a general theory on the sustainability of federations. Two reasons why federations persist are offered, both of which are premised on the assumption that threats are no longer imminent and that federalism will always produce winners and losers among the constituent states. Given the nature of federal institutional arrangements, it must always be a minority or minorities that are the winners in relation to the majority. First, federalism may be maintained because it is simply too inconvenient to abandon it. In other words if the costs of the alternatives (dissolution, or making the transition to unitary government) are higher than those associated with the status quo then it will be maintained. This claim is evidently true. The second reason for the maintenance is more interesting. Federations may persist because the costs to the majority of maintaining it are slight. If the system permits a majority to overrule the minority when the majority opinion is strongly held, then the costs of federalism are likely to be low. Just this happened during the 1960s when Northern opinion against the racism of the southern states was sufficiently strong to force the South to change its political culture and institutional practices. But, in the modern context, at least, Riker fails to develop this point by linking the dynamic to institutional arrangements. In the American case it was a nationalising and ascendant Democratic Party that was able to overcome historically established veto points in Congress that facilitated the civil rights revolution. But what if the institutional vetoes were stronger (or were constitutionally enshrined) or the strength of opinion among the Southern states was sufficient to continue resistance? The latter was certainly the case in 1861 and the former applies in a number of federal systems whose design has deliberately been informed by the need to protect territorial minorities. Such is the case with the European Union and with Switzerland. What is it that sustains such systems while others are constantly under threat of dissolution (Canada, India, Nigeria) or actually dissolve (the Soviet Union, Czechoslovakia, Yugoslavia)?

While Riker’s framework is insufficient to answer these questions, it may be possible to do more than simply construct taxonomies or tell interesting stories in theoretically informed ways. Thus, Filippov, Ordeshook and Shvetsova advance understanding beyond Riker's intuitions by synthesizing game theoretic and institutionalist approaches in an attempt to construct a general theory on the self enforceability of federal institutions. Their intention is to set out the theoretical conditions that will best guarantee long term compliance with constitutional ground rules. These include the specification of those constitutional designs that will best represent the balance of national and provincial interests by institutionalizing bargaining to ensure compliance (Filippov, Ordeshook and Shvetsova, 2001, Chapters 4 and 5). Institutions must be designed in such a way as to make it in the self-interest of regional elites to co-operate with national elites. National elites should, in turn, have self-interest in protecting the interests of all citizens in all of the constituent states. The resulting coalition building, bargaining and negotiation between different levels of government will be a continuing process. A relative loser at any point in the bargaining episode will know that the rules of the game are such that an opportunity to recover this loss will be available at a later point in the game (Ordeshook and Shvetsova, 2000, 31-33). Regional interests will, therefore, have little or no incentive to defect or to secede. And national interests will have little incentive to usurp regional power through excessive centralization.

But which institutional design is likely best to facilitate such a dynamic? Most scholars working in this area invoke the American experience, in part because it has survived in a functioning form over a very long period (the Civil War apart), and in part because we know a great deal about it. In order to demonstrate it’s favorable features US federalism it is often contrasted with Canadian. While space considerations prevent a full discussion of this theme, it is possible to infer from the work of Riker and others three institutional features that have helped facilitate the survival of federations. What follows is a brief discussion of these features in comparative context. In particular, an attempt is made to produce examples from other systems and in particular Switzerland and the European Union (EU) that show how any general theory based on inferences from the American experience need to be carefully qualified.

1. The Role of Upper Houses. The reasoning here is simple and familiar: successful federation is aided by the presence of an upper house based on territorial representation. Bicameralism facilitates the representation of state interests at the national level and permits coalition building and bargaining both between states and between state and the central government. Canadian federalism with its politically appointed and feeble upper house denies the provinces these opportunities. The American Senate provides them Ordeshook and Shvetsova, 2000). There may, additionally be some benefit from delegated as opposed to directly elected upper house members on the assumption that. delegated representatives will more likely serve the interests of regional legislatures.

But among established federations it is difficult to find examples of upper houses that act as a true 'chambers of the states' whether members are directly or indirectly elected. As Riker showed of the US Senate the 'doctrine of instructions' that bound Senators to the directives of the state legislatures was always weakly enforced and quickly passed from the scene (Riker 1955). Even the claim that bicameralism served to broker the interest of the South in the 1880-1970 period is probably unsustainable. Such interests were served by the one party nature of the region that allowed the manipulation of parliamentary rules in both houses and by an electoral system that obliged Democratic presidential candidates to build winning coalitions with Southern Democratic party organisations. More recently Senators, along with upper house members in other countries have 'gone national.' In the Swiss Standerat, for example, which is constitutionally equal to the lower house and whose members are elected on a territorial rather than population basis, members take broad ideological positions on national issues rather than serve individual cantons or groups of cantons - and this in spite of the veto that a small number of cantons in the chamber can exercise over national policy.[8] (Linder, 1998, Chapter 2). Instead, the cantons’ interests are served by other institutional arrangements including canton level referenda and initiatives (of which more later). Even the delegated membership German Bundesrat only occasionally acts as a forum for brokering central/Lander conflicts. Most studies confirm that national party cues are more important than territory in determining member voting (Jeffrey 1999; Lehmbruch, 1978, 151). Of course whether in the Bundesrat or in other institutions, party may play crucial self-sustaining roles for the German polity (Lehmbruch, 1989; Silvia, 1999). But so it does in other political systems irrespective of whether they are federal or unitary. Only the EU Council of Ministers is a true chamber of the states but it operates largely according to unanimity decision rules that guarantee powerful limits on central power. Should Calhoun's concurrent majority have been adopted in the ante bellum US, the result would have been the same - the peripheralization of the federation. When upper houses operate according to plurality decision rules they often end up as co partners in a national bargaining game rather than as national/regional facilitators. And this applies to both directly and delegated upper house members. Historical experience suggests, therefore, that although territorially based upper houses may sometimes help broker national and regional interests in ways that help sustain federations, they are not the pivotal institutions in this process.

2. Liberalism v. Populism? When specifying the optimal rules of constitutional design and the principles that should apply to the bargaining undertaken by political elites, Riker (following Madison) shows that left unchecked by institutional constraints, the popular will puts democracy at risk. A functioning federation (or indeed any democracy) would require inter alia multicameralism, a division of powers between national and regional government and an independent judiciary in order to check the possibility that the popular will go unchecked and lead to oppression and tyranny (Riker, 1982). It is easy to infer from this axiom that populistic devices such as initiatives and referenda are inimical to the Madisonian notion of democracy. This certainly would be the case with Mussolini style executive led plebiscites. But the Swiss experience shows that placed in the appropriate institutional context, referenda and initiatives can actually help sustain federation. Swiss democracy has evolved over several hundred years specifically to accommodate territorially defined linguistic and religious difference. Sustaining the confederation is a continual process based on a number of related informal and formal institutional devices, all of which are designed to contain central power and preserve regional autonomy. One is the acceptance among elites that informal negotiation, bargaining and coalition building must precede formal institutional approval. This happens at all levels, municipal, cantonal and national (Kobach, 1997; Armingeon, 2000). Elites are obliged to pursue such a strategy because a winning policy has to jump three enormous hurdles before it can succeed. First the institutional veto points are numerous. These include, bicameralism, legislative review (vernehmlassungsverfahren[9]), administrative devolution, judicial review and a collective executive). Second, at any time and at any level a proposal can be stopped by referendum. In such a context the incentive to build winning coalitions through long and protracted bargaining and negotiations is enormous. As one observer puts it 'Probably the strongest incentive for co-operative behavior stem from direct democracy '(Armingeon, 2000, 125). But there is a third constraint on policy makers - for virtually every government function at every level is enshrined in the 196 articles of the Swiss constitution This includes an extraordinarily specific delineation of central, cantonal (and local) powers all the way down to particular levels of taxation. The constitution can in turn be amended only by the famous double majority vote - a national majority and a cantonal majority. Thus the population asymmetry of the cantons gives to the smallest 11.5 cantons representing just 14% of the population a theoretical veto over constitutional amendments (Although typically the figure is around 24%, Linder 1998, Table 2.5). Given the specificity of the constitution, amendment attempts are not infrequent and success is quite rare. Hence the Swiss system structures incentives in such a way that coalition government is the norm (almost the rule) at every level. Political parties are 'coalitions of cantonal parties which in turn are coalitions of communal parties' (Steinberg, 1996, 113; see also Gruner and Pitterle, 1983). The institutional rules discourage sectional parties, Canada and (historically) US style, and encourage co-operation and sustainability. Central power is limited (the Swiss have almost the lowest GDP in the public sector among OECD countries), and the federal government taxes and spends significantly less than cantonal and communal governments. At the same time, no one canton has an incentive to leave. As Linder puts it: 'Changing the rules of federalism is a game to be played under the existing rules of federalism, and there is no reason for minorities to renounce their long-held minority rights when asked to do so' (Linder 1998, 76). So because of rather than in spite of direct democracy, the Swiss probably come closer to meeting the Madisonian ideal than any other federation.

3. The Structure and Functioning of the Party Systems. Riker ascribes to the structure of the political party system a special role in brokering conflicts between center and periphery. So, while underlying social conditions are the building blocks of the party system, how - and indeed whether - the party system accurately reflects the balance of localist and centralist interests will depend largely on constitutional arrangements. Hence:

‘In a variety of governments, then, the structure of the parties parallels the structure of federalism. When parties are fully centralized, so is federalism (e.g. in the Soviet Union and Mexico). When parties are somewhat decentralized, then federalism is only partially centralized. Because of this perfect correlation of, at least the two extreme categories of federalism with party structure, the inference is immediate: One can measure federalism by measuring parties. The structure of parties is thus a surrogate for the structure of the whole constitution (Riker 1975, 137).

Riker (with collaborator Schaps) constructs an index of disharmony that measures the extent to which national parties in government control constituent parties and governments and whether or not party discipline exists at the national level. The authors conclude that the index of disharmony is effectively an index of decentralization both of the party system but also of the federal constitutional structure.[10] But this measure is too unsubtle. It places centralized federations such as Australia and Germany in the same category as decentralized federations such as Switzerland. All are characterized by having the same parties operating at all levels (with minor exceptions) and all have relatively strong party discipline at the center. The same parties may operate at all levels in Switzerland (also with minor exceptions), but they are highly decentralized. Common party labels are retained because in order to win national representation, influence national policy and avert referenda; it is necessary to build coalitions at every level and in every institutional context. This is facilitated by the organisational connective tissue - however weak - that comes with shared identification (McKay, 2001 and sources cited). Party systems may be decentralized organizationally or ideologically but the same parties may operate at all levels. The extreme case of a decentralized party system in the Rikerian sense, is the European Union (Figure 1) where all the parties are regional and all are decentralized.

[Figure 1 here]

Swiss and American style coalitional parties are more likely to evolve in decentralized federations whose institutional design facilitates central/regional bargaining and accommodation. We have already discussed the Swiss case, and a large literature on the United States confirms how the electoral system, the separation of powers and checks and balances have encouraged bottom up coalition building that has, since the Civil War at least, helped accommodate Southern interests and thus ensured political stability (see for example, Alston and Ferrie, 1999).

If institutional arrangements encourage regional parties based on strong regional identities (as in Canada) then the dangers of break down and secession are greatly increased. But whether such parties develop depends in turn on the level and scope of central government activities. In the Canadian case as long as the scope of federal activities was limited regional parties were weak. During the mid 20^th Century however, a succession of centralizing federal governments provoked regional interests into using provincial legislatures as vehicles for the creation of distinctively regional parties. In 1900 national parties controlled all the provincial governments. By 1950 40% were controlled by regional parties (Dunn, 1996, Appendix 1). It is not obvious that this development could have been predicted. In 1867 many of the framers of the Canadian constitution expected the system to evolve into a unitary state. They thus adopted centralist parliamentary institutions that facilitated little in the way of national/regional linkages. The French speaking population was small and geographically isolated in Lower Canada. Fifty years later it had grown substantially in size while retaining its linguistic and cultural identity. Once the scope of federal power over the Francophone community had increased and denied access to the anglophone political center, they were obliged to create separate institutions through distinctive regional political parties. In order to facilitate bargaining and maintain the union a number of ad hoc coordinating devices have evolved including the First Ministers Conference. It is generally agreed that the original failure to ‘design in’ representation of the provinces in central government decision making has produced an abrasive, confrontational and unstable polity (For a discussion see McRoberts, 1997).

But we do not know how the balance of regional loyalties will change over time. Nor do we know how this will relate to changing balances in the level and scope of central as opposed to regional power. Who, just 15 years ago, would have predicted the successful adoption of monetary union in the EU, for example, which has greatly increased the potential scope of federal power? This major policy change has been implemented in the almost complete absence of mediating institutions, even though it followed months and years of bargaining among elites in the participating countries. In this context, the dangers of an eventual breakdown or of secession remain real (McKay, 1999).

Conclusions

Riker's insights into why federations emerge and what sustains them over time continue to provide us with the most effective heuristic tools at our disposal in this subject area. His characterization of federation as a result of a bargain between central and regional elites intent on averting external or internal threats has yet to be shown to be empirically false. Only his insistence that the threat be military or diplomatic needs to be qualified to include economic and cultural threats. Delineating the conditions for the maintenance of the bargain once the threat has receded is a much more challenging proposition. A priori it is plausible to assume that the presence of particular institutional arrangements would be necessary to ensure compliance with the rules of the game and mediate central/ regional disputes. But generalizing in this area is problematical. Territorially based upper houses often play little or no role as champions of state interests. Under the right circumstances, direct democracy may serve the cause of federal stability, as may highly codified and inclusive constitutions. Under other circumstances they may not. The role of political parties is equally complex. Whether they help maintain the federation or not would seem to depend on a subtle interaction of institutional arrangements with the strength of regional loyalties and the policy scope of federal governments in relation to state governments. The most challenging problem is that, whether at the inception of federalism or at any later point, we do not know either how the balance of loyalties or the balance of federal/local responsibilities will change over time. What looked like an appropriately centralist Canadian constitution in 1867, was inappropriate 75 years later.

This paper has provided only limited examples both by country and by institution. If extended, the project becomes even more formidable. How, for example can constitutions be designed to accommodate primary state dominance (inter-state asymmetry in size and power)? Or which design is appropriate when ethnic, religious or linguistic minorities are not neatly separated by territory?

However, while any attempt to produce a general theory by specifying the necessary and sufficient conditions for sustaining federations over time, is almost certainly a chimera, Riker's insights into the nature and functioning of federal government are enormously helpful in generating working hypotheses that can guide research (see Fillipov, Ordeshook and Shvetsova, 2001, Chapter 5). For example we can easily see - although admittedly with the benefit of hindsight - how an informed social choice framework would have led to the creation very different constitutional arrangements to the British imposed federal settlement in such countries as Nigeria. A disastrous civil war might have been averted. Institutions matter. The trick is identifying the appropriate institutions in the right place at the right time.

References

Alston, Lee J. and Ferrie, Joseph P. (1999), Southern Paternalism and the American Welfare State: Economics, Politics, and Institutions in the South, 1965-1965, Cambridge and New York, Cambridge University Press.

Armingeon, Klauss (2000), Swiss Federalism in Comparative Perspective, in Ute Wachendorfer-Schmidt (Ed), Federalism and Economic Performance, London, Routledge, 2000, 112-129.

Birch, Anthony H. (1966), Approaches to the Study of Federalism, Political Studies, 14, 15-33.

Denoon , Donald and Wyndham, Marivic (2000). 'Australia and the Western Pacific,' in Andrew Porter (Ed), the Oxford History of the British Empire. Vol. III, The Nineteenth Century, Oxford, Oxford University Press, 546-572

Deutsch, Karl, et al (1957), Political Community in the North Atlantic Area, Princeton, NJ, Princeton University Press.

Dikshit, Ramesh Dutta (1971) 'Military Interpretations of Federal Constitutions: A Critique,' Journal of Politics, 33, 180-189.

__________________, (1975), The Political Geography of Federalism: An Enquiry into Origins and Stability, New Delhi, Macmillan.

Dunn, Christopher (1996), Provinces: Canadian Provincial Parties, Peterborough, Ontario, Broadview Press.

Dyson, Kenneth (1994), Elusive Union: The Process of Economic and Monetary Union in Europe, London, Longman

__________________, and Featherstone, Kevin (1999), the Road to Maastricht: Negotiating Economic and Monetary Union, Oxford, Oxford University Press.

Friedrich, Carl. J. (1968), Trend of Federalism in Theory and Practice, London, Pall mall Press.

Fillipov, Mikhail, Ordeshook, Peter C. and Shvetsova, Olga (2001), 'Designing Federalism: A Theory of Self-Sustainable Federal Institutions', Manuscript, Washington University, St. Louis.

Galligan, Brian (1995), A Federal Republic: Australia's Constitutional System of Government, Melbourne and Cambridge, Cambridge University Press.

Gruner, Erich and Pitterle, Kenneth J. (1983), 'Switzerland's Political Parties,' in Howard R. Penniman (Ed), Switzerland at the Polls: The National Elections of 1979, Washington DC, American Enterprise Institute, 30-59.

Jeffery, Charlie and Savigear, Peter (Eds) (1991), German Federalism Today, Leicester, Leicester University Press.

Jeffery, Charlie (1999), 'The Role of the Bundesrat,' in Brzinski, Lancaster and Tushhoff (Eds), Compounded Representation in Western European Federations.

Jennings, Ivor (1940), A Federation for Western Europe, Cambridge, Cambridge University Press.

King, Preston (1982), Federalism and Federation, London, Croom Helm

Kobach, Kris W. (1997), ‘Spurn they Neighbour: Direct Democracy and Swiss Isolationism,' West European Politics, 20, 185-211.

Lehmbruch, Gerhard (1978), 'Party and Federation in Germany: A Developmental Approach,' Government and Opposition, 13.

Lehmbruch, Gerhard (1989), 'Institutional Linkages and Policy Networks in the Federal System of West Germany,' Publius, 19, 221-235.

Lemco, Jonathan (1971), Political Stability in Federal Governments, New York, Praeger.

Linder, Wolf (1998), Swiss Democracy: Possible Solutions to Conflict in Multi Cultural Societies, London and New York, Macmillan, St. Martin's Press.

MacKay, R.W.G. (1940), Federal Europe, London, Michael Joseph.

McKay, David, (1999a), 'The Political Sustainability of European Monetary Union,' British Journal of Political Science, 29, 463-485.

________________ (1999b), Federalism and European Union: A Political Economy Perspective, Oxford and New York, Oxford University Press.

_______________, (2001), Designing Europe: Comparative Lessons from the Federal Experience, Oxford, Oxford University Press.

McMinn, W. G. (1994), Nationalism and Federalism in Australia, Melbourne and Oxford, Oxford University Press.

Manor, James (1998), 'Making Federalism Work', Journal of Democracy, 9, 29-36

McRoberts, Kenneth (1997), Misconceiving Canada: the Struggle for National Unity, Toronto and Oxford, Oxford University Press.

Malcolm, Noel (1995), 'The Case Against Europe,' Foreign Affairs, March/April.

Mearsheimer, John C. (2001), The Tragedy of Great Power Politics, New York, Norton Mitra, Subrata K. (2000), The Nation, State and the Federal Process in India, in Ute Wachendorfer-Schmidt (Ed), Federalism and Political Performance, London, Routledge, 2000, 40-57.

Moore, Robin J. (2001), 'Imperial India, 1858-1914,' in Andrew Porter (Ed), The Oxford History of the British Empire, Vol. III, The Nineteenth Century, Oxford, Oxford University Press, 422-446.

Moravscik, Andrew (1998), The Choice for Europe: Social Purpose and State Power from Messina to Maastricht, Ithaca and London, Cornell University Press.

Ordeshook, Peter C. and Shetsova, Olga (1997), 'Federalism and Constitutional Design,' Journal of Democracy, 8, 27-42

Quick, J. and Garran, R.R. (1901), Annotated Constitution of the Commonwealth of Australia, Sydney, Angus and Robertson.

Riker, William H. (1955), 'The Senate and American Federalism,' American Political Science Review, 49, 452-469

_____________, (1964), Federalism: Origins, Operation, Significance, Boston, Little, Brown.

______________, (1975), 'Federalism,' in Fred I. Greenstein and Nelson Polsby (eds), The Handbook of Political Science, Volume V: Government Institutions and Processes, Reading MA, Addison Wesley.

______________, (1982), Liberalism Against Populism: A Confrontation Between the Theory of Democracy and the Theory of Social Choice, Illinois, Waveland Press

______________, (1987), The Development of American federalism. Boston, Kluwer.

______________, (1996), 'European federation: Lessons of Past Experience,' in Joachiim Jens Hesse and Vincent Wright (Eds), Federalizing Europe: The Costs, Benefits and Preconditions of Federal Political Systems, Oxford, Oxford University Press, 1996, 9-24.

______________, and Schaps, Ronald (1957), 'Disharmony in Federal Government,' Behavioural Science, 2, 276-90.

______________, and Lemco, Jonathan (1987), 'The Relation Between Structure and Stability in Federal Governments, in Riker (1987), 113-134.

Silvia, Stephen J. (1999), 'Reform Gridlock and the Role of the Bundesrat in German Politics,' in Brzinski, Lancaster and Tuschhoff (Eds), Compounded Representation in Western Europe, pp. 167-181.

Steinberg, Jonathon (1996), Why Switzerland? Cambridge and New York, Cambridge University Press.

Tiebout, Charles M. (1956), 'A Pure Theory of Local Expenditure,' Journal of Political Economy, 74.

Washbrook, D.A. (2001), 'India 1880-1860: The Two faces of Colonialism' in Andrew Porter (Ed), the Oxford History of the British Empire: Vol. III, The 19^th Century, Oxford, Oxford University Press, 295-421.

Watts, Ronald L. (1996), Comparing Federal Systems in the 1990s, Institute of Intergovernmental Relations, Queens University, Ontario.

Organisational and ideological centralization

High

Low

High

Spatial

unity

Australia

Canada 1900

Germany 1970

Germany

Canada

Switzerland

European Union

Low

Figure 1. The territorial dimension to party systems in five federations and the European Union

Source: McKay 2001, Figure 8.1