Party System Change

Peter Mair. Handbook of Party Politics. Editor: Richard S Katz & William Crotty. Sage Publications, 2006.

Introduction: Party System Change and the Classification of Party Systems

The key problem with the phenomenon of party system change is that it is seen as either happening all the time or as scarcely happening at all. This is not a reflection on the state of the world, and it is not intended to suggest that the frequency of cases of party system change echoes the proverbial complaint about London buses: you wait for ages for one to come along, and suddenly three arrive together. Rather, it is a reflection on the approaches to the classification of party systems that are currently to be found in the literature, and that have already been discussed by Steven Wolinetz in Chapter 6 of this volume.

There are two approaches to classifying party systems that are relevant here, in the sense that both have quite distinct implications for how party system change is treated. On the one hand, there is the traditional comparative approach developed by scholars such as Duverger, Blondel, and Sartori, which aims to categorize party systems into distinct classes or types, such as two-party systems, systems of moderate pluralism, multiparty systems, or whatever. Following this approach, party systems scarcely change at all, in that a change of party system necessarily involves the case in question moving from one category to another—from the two-party to the multiparty category, or from moderate pluralism to polarized pluralism, and so on—and the conditions that allow for such a reclassification are usually so demanding that it rarely occurs in practice. Some years ago, for example, I sketched an analytic history of the Irish party system in which, by using Sartori’s categories, I tried to show that there had been a shift from polarized pluralism to moderate pluralism, and then to a predominant-party system over a period of some 20 years (Mair, 1979). In a similar vein, Arian and Shamir (2001: 705) have recently pointed to what they see as evidence of ‘tremendous change’ in the Israeli party system since 1948: the transition from a stable dominant-party system to a competitive two-bloc system and then to the contemporary situation in which the configuration is highly unstable. These sorts of shift seem quite exceptional, however. More typical is the British case, which, over a much longer period, disallowing short-term flux and wartime peculiarities, scarcely deviated from its well-entrenched two-partyism. To see party systems in terms of discrete categories is therefore to bias one’s analysis in favour of the absence of change.

The other approach to classifying party systems effectively avoids the issue of classification entirely, and instead employs continuous numeric variables to ‘summarize’ or ‘define’ the party system, usually for the purposes of cross-national inquiry. These continuous variables are almost always based on a calculation of the number and relative size—whether electoral or parliamentary—of the parties present in the system. In the earlier literature, the preferred version was Rae’s (1968) index of fractionalization (see Shamir, 1984), whereas the more recent literature tends to prefer the version of this index that is modified as ‘the effective number of parties’ (Laakso and Taagepera, 1979; see also Wolinetz, Chapter 6, this volume). Following this approach, differences between party systems are treated as a matter of degree rather than of kind, and hence party system change is a continuous phenomenon. It is also a confusing phenomenon, since these particular summary measures conflate changes in the number of parties with changes in their relative weights, thereby leaving the observer in the dark as to what sort of change was actually involved (see Pedersen, 1980). For example, the British party system could be said to have ‘changed’ when the effective number of parties fell from 3.45 following the election of 1983, to 3.33 following the election of 1987 (see the figures in Webb, 2004: 22), even though most observers, and, indeed, most British citizens, would have been very hard pressed to identify what precisely had happened to account for this shift. To see party systems in terms of these continuous variables is therefore to record constant and confusing change, and to bias one’s analysis against the identification of stability.

Over the past three decades or so, the literature on party systems has tended to move away from the discussion of discrete categories and to rely more heavily on continuous variables. In part, this is because there has been very little new thinking on how to classify systems since the seminal work of Sartori (1976; see also Wolinetz, Chapter 6, this volume). In part, it is because of a drift away from case-sensitive and thickly descriptive comparative case studies towards the analysis of more broad-ranging cross-national research questions. Categorical classifications of party systems do not easily lend themselves to quantitative research; or at least they are not likely to result in attractive correlation coefficients. Lijphart’s influential study of alternative models of democracy is a very good case in point here. Coming from the older comparative politics tradition, in which he has also been one of the leading figures, Lijphart (1999: 62-9) begins his discussion of the relevance of party systems to his models of democracy by highlighting the categorical distinction between two-party systems and multiparty systems, and by linking the former type to the majoritarian model of democracy and the latter to the consensus model. As his analysis progresses, however, and as he begins to apply his framework, this categorical distinction becomes translated into a continuous measurement of the effective number of parties, eventually leading to the calculation of a distinct numerical score for each of the 36 polities that concern him. This is the approach that has now come to dominate quantitative cross-national research. By affording room for measuring endless variation between party systems, however, and by failing to establish plausible thresholds that could be used to identify the emergence of a new or a different party system, it implicitly renders meaningless any notion of party system change.

Against Numbers

It also renders meaningless any notion of party systems. These summary variables count the number of parties in a polity and take account of their relative size, but, in itself, this tells us little of importance about the system as such, or about system change. Indeed, with the exception of the limiting case of polities that have only two more or less equally sized parties, and hence that maintain a pure two-party system, numbers as such have little systematic relevance. We speak of Britain, or the USA, or Greece, or especially Malta, as having two-party systems, but, as Wolinetz reminds us, this is not always because there are only two parties in each of these polities (see also Mair, 2002b: 93-4). Indeed, there were 12 parties that actually won representation in the British House of Commons in 2005—some of them, to be sure, could have travelled there in the back of a taxi—as did one independent candidate. In Malta, however, which has the purest and most fully mobilized two-party system in Europe, the Democratic Alternative—which was a Green party and the third party of the system—won only 0.7 per cent of the vote on a 96 per cent turnout in 2003, and thereby scarcely rocked the established two-party balance. In any case, the more general rule is that two-party systems are two-party systems not because there are only two parties as such, but because we judge that there are only two parties which are involved in, or are relevant to, government formation. It then also follows that multiparty systems are multiparty because more than two parties are involved in government formation. But beyond possibly providing this information, knowledge of how many parties exist in the polity can tell us next to nothing in itself about how the party system works. For that, we need to know how the various parties can and do act.

This stricture also applies to the classic Sartori typology, despite Sartori’s own best efforts to rescue party numbers as a key variable. For Sartori (1976: 128), the number of parties matters, in that the format of the system has ‘mechanical predispositions’—that is, knowing how many parties exist gives the observer a good indication of how these parties are likely to interact. But even within Sartori’s own framework, this is true only to a certain limited extent. Numbers as such cannot allow us to distinguish the different mechanisms that operate within the many and varied cases of moderate pluralism, for example, in that they cannot tell us whether such a system is likely to be characterized by competing coalitions and wholesale alternation in government, or by overlapping coalitions and partial alternation. Moreover, while numbers can be important in marking the crucial difference between moderate and polarized pluralism, this is not always the case, particularly when the fragmentation in question has been induced by a multiplicity of domains of identification rather than by a stretching of the dimension of competition, that is, by polarization (Sani and Sartori, 1983: 335-7). Numbers alone can also give us no indication about whether we are dealing with a predominant-party system.

Since numbers alone have little meaningful systemic relevance, numerical change cannot be considered to be the same as system change. Hence, the fact that the number of parties has grown significantly through the postwar years in the advanced industrial democracies (Dalton et al., 2000: 40-3; Mair, 2002a: 133-5) does not necessarily indicate that the party systems in these democracies have been transformed. Denmark offers one of the most telling examples in this regard. Until the end of the 1960s, in the context of what had become ‘one of the most dull countries to deal with for an empirically oriented student of voting behaviour’ (Pedersen, 1968: 253), the party system had contained some five parties of varying electoral strengths. In 1973, this number suddenly doubled as the result of a so-called ‘earthquake’ election, and since then there have always been at least eight parties represented in the parliament (Bille, 1989; Bille and Pedersen, 2004). In terms of the functioning of the system itself, however, and in terms of the way parties interact with one another, it can be argued that even post-earthquake Denmark reflects more evidence of persistence than of change (Mair, 2002b: 101-2). In this sense, if somewhat ironically, we see that numbers do not always count.

Reverting to the older classifications of comparative politics also offers little guidance in the identification and interpretation of party system change, however. The most traditional classification, that which distinguishes two-party and multiparty systems, is simply too crude for most scholarly purposes. There are not enough two-party systems in the world to allow this to function as a balanced classification, and the multiparty category is itself too overcrowded and undifferentiated. To limit instances of party system change to shifts across this particular boundary would be to impoverish our understanding. The same holds true when looking at Sartori’s typology which, however insightful, is also proving less and less appropriate to the task of identifying distinctions within the contemporary world. Indeed, the demise of the traditional anti-system party (see also below) effectively brings the range of Sartori’s types back to the two- versus multiparty distinction, although in his more nuanced version this distinction translates as that between two-party systems and systems of moderate pluralism.

Understanding Party System Change

In order to understand and identify the phenomenon of party system change, we therefore need to adopt an alternative perspective, one that is neither overly constrained by the traditional classifications, on the one hand, nor too easily dissolved into crude quantitative indicators, on the other. This alternative perspective also needs to address the essence of the party system; that is, it needs to address the principal modes of interaction between the parties and the way in which they compete with one another (see Smith, 1989a, 1989b). Party system change at the margins, even if this could be specified, is of little interest. What matters is change at the core.

Let us be clear on the terms of reference here. First, the core of any party system qua system is constituted by the structure of competition for control of the executive. Despite the differing perspectives advanced by the various classifications to be found in the literature, ranging from those of Duverger, Dahl and Rokkan, to those of Blondel and Sartori, this remains a point of more or less widespread agreement: defining a party system begins with an understanding of how governmental power is contested. It is here that the core of the party system is to be found, and hence the parties which count are those that are involved in or have an impact on that competition.

Second, it then follows that a party system changes when there is a change in the structure of competition. And this, in turn, may be broken down into three related components. The first of these identifies change in the structure of competition as taking place when there is a change in the prevailing pattern of alternation in government, such that, for example, there is a shift from a prevailing pattern of wholesale alternation to one of partial alternation, or vice versa, or there is a shift from a prevailing pattern of non-alternation to one of partial or wholesale alternation. The second component refers to the extent to which the governing alternatives in the system prove stable or consistent over time, or whether they involve innovative formulae. The important change in the structure of competition that is identified here occurs when a period of consistency in the make-up of government is then succeeded by a new and innovative alternative. The third component refers to the question of who governs, and to the extent to which access to government is either open to a wide range of diverse parties or limited to a smaller subset of established governing parties. Seen in this light, change in the structure of competition is perhaps most easily observed when it involves a new party arriving in office; by definition, this will also involve the adoption of an innovative governing formula.

This also allows us to classify party systems as such, of course, for by combining information about the patterns of alternation, the degree of innovation, and the access of new parties to government we can begin to distinguish between closed and open structures of competition, and hence between different party systems. At the same time, we can also gain a sense of the degree of systemness of any individual party system. The two limiting cases here are, on the one hand, those party systems in which competition is wholly closed, and in which the pattern of alternation is entirely predictable, the competing protagonists are wholly familiar, and no new party or alliance has any real hope of gaining government; and, on the other hand, those in which competition is wholly open, in which there is little that is predictable or familiar in either the patterns of competition or the make-up of the competing forces, and in which new parties and alliances need place no limit on their expectations. The one case reflects an exceptionally high level of systemness and constitutes a very strong party system, while the other reflects a very low level of systemness and a very weak party system. Indeed, at the latter limit, we see what is scarcely a party system at all, since there is so little here that is clearly patterned: the parties themselves are very fluid, and their interactions relatively shapeless. In their application of a similar approach to the party systems of Latin America, Mainwaring and Scully (1995a) adopt the useful terms ‘institutionalized’ systems and ‘inchoate’ systems to refer to this distinction; strictly speaking, however, an inchoate system, or a system with a wholly open structure of competition, is not really a system at all. Systemness implies institutionalization.

More importantly for present purposes, however, these three elements afford a practical set of guidelines for identifying when and how party systems change. As specified above, a party system changes when there is a change in the prevailing structure of competition. That is, a party system changes when there is a change in the pattern of government alternation, when a new governing alternative emerges, and/or when a new party or alliance of parties gains access to office for the first time. It follows that for any such change to occur and be noted, there must have already existed a prior and well-established pattern of competition, and the importance or weight of any change will depend on how stable and well established that prior pattern proved to be.

This last may seem an obvious point, but it is worth spelling out more fully: under normal circumstances, we cannot identify a new mode of alternation in government except by contrast to an already existing pattern, and we cannot speak of an innovative governing formula except by reference to other formulae that are already established. In other words, we cannot speak of a change of party system unless a more or less robust system was already in place. There is one important exception to this rule, of course, and that is when an otherwise ‘inchoate’ system is replaced by a more patterned system, as happens when newly democratized systems become established for the first time (Holmes, 1998; see also Sartori, 1994: 37). In this case there is obviously no older established pattern that is being displaced, but instead, as has recently been the case in the Czech Republic, for example (Kopecký, 2005), a formerly incoherent set of interactions begins to take shape and acquires structure. But although this is, formally speaking, a process of party system change, it might be better and more informative to treat it as a case of party system institutionalization. At the other extreme, we can also conceive of cases of party system deinstitutionalization—that is, when a prevailing structure of competition breaks up or collapses, and is succeeded by a unformed, inchoate set of interactions. In this case, an existing party system loses shape and becomes destructured without any alternative system emerging in its place. These extremes notwithstanding, the majority of cases of party system change involve the replacement of one type by another. That is, a prevailing structure of competition is broken, and is replaced by a new structure of competition. Just such a change appears now to be the case in contemporary Germany, for example, in that a new pattern of bipolar competition—pitting a centre-left coalition of Social Democrats and Greens against a centre-right coalition of Christian Democrats and Liberals—seems set to replace the older pattern of overlapping opposition, whereby Christian Democrats and Social Democrats sought to gain office with the help of the once pivotal Liberal party. A similar shift towards bipolarism appears to be occurring in Austria. In Italy, the long-standing postwar party system, most aptly characterized in Sartori’s terms as a system of polarized pluralism, in which government was always formed from the centre, has also moved towards bipolarism, with the emergence of two more or less coherent and quite distinct electoral alliances that now dominate national competition (Bartolini et al., 2004). All of these shifts are easily identified by reference to the changing patterns of alternation in government, and to the ways in which executive office is being contested. More importantly, perhaps, they have also all involved a break with what had been a persistent and longstanding pattern of competition, even though it may take some time before we can be sure that a new stable structure has been established in place of what went before. Indeed, as Müller and Fallend (2004: n. 6) note, verifying that this sort of change has taken place needs patience.

This is far from being a uniform or standardized process, however, and any application of these guidelines requires sensitivity to the peculiarities of the case. This may also preclude an easy application within cross-national quantitative research. Not all cases of new governing alternatives or even of new parties arriving in government will constitute cases of party system change, for example, and the impact of such changes will clearly differ from one polity to the next. In other words, while the factors identified above offer a useful set of common guidelines, their weight will always remain context-specific. In one context, the structure of competition might be normally very open and flexible, and we might only be able to speak of the emergence of something new at the point when the system closes down. In another, regular access of new parties into government might constitute one key feature of the party system, such as it is, and we might only begin to talk of change when this pattern ceases to apply. In other words, adding a new party to the ranks of the governors will not make for a new party system if the provision of access to new parties has always been a normal part of the political process.

The contrasting cases of Finland and Ireland offer useful reference points in this regard. Across the past half-century of democratic politics, Finland experienced more than 30 changes in the partisan composition of government, almost all of which resulted from instances of partial alternation—that is, from a reshuffling of existing coalitions and/or from the incorporation of additional parties into preexisting alliances. For more than two-thirds of this period, Finland was governed by innovative governments, that is, by parties or combinations of parties that had never previously held office using the same governing formula (Mair, forthcoming). The result was a remarkably unstructured and unpredictable pattern of competition, and hence a remarkably open and poorly defined party system, such that the factors that would be seen to constitute a case of party system change in a more closed system proved much less relevant in the Finnish case. In Ireland, by contrast, these sorts of changes were anything but marginal or routine, and when the newly formed Progressive Democrats took office in 1989 as the junior coalition partner of the long-dominant Fianna Fáil party, it marked a major change in the system. For the first time in its long and successful history, Fianna Fáil took part in a coalition government, and for the first time since at least 1948 competition no longer revolved around the opposition between Fianna Fáil on its own, on one side, and a coalition of more or less all remaining parliamentary parties, on the other. In the Irish case, in other words, a very long-standing and familiar pattern was broken open by the sort of innovative coalition that would have proved quite commonplace in Finland.

The Irish case is also a useful illustration of how party systems stabilize around particular assumptions, and how they can suddenly be broken open when those assumptions are challenged. The simple question of time is obviously important here, in that the cumulating daily practice of politics may lead both voters and party leaders to become used to thinking within a particular, and hence institutionalized, set of terms of reference. If the range of governing alternatives has been limited in the past, then this is likely to encourage both observers and participants to believe that they may also be limited in the present. If, to cite this Irish example, previous governments have been formed only by either Fianna Fáil on its own, on the one hand, or a coalition of more or less all other parties, on the other hand, then it is unlikely that voters will be easily persuaded to think in terms of any alternative constellations. It is in this sense that a system becomes predictable and familiar: the alternatives appear to be constrained by particular options being ruled out as unthinkable. Constraining the alternatives will also be the result of elite choices and elite political culture, with the leaders of the established parties being keen to promote the maintenance of the patterns that have served to guarantee them success in the past (see Schattschneider, 1960: 60-74). Conversely, party systems may change when new leaders or new parties begin to explore and then act upon new alternatives.

The nature of the wider institutional structure within which the party system is located can also serve to limit the scope for party system change. In the first place, this wider institutional context will help to define and hence to limit the potential alternatives which are seen to be available. As noted above, bipolarity in the party system of the Fifth French Republic was clearly helped by the institution of the presidency, for example, and by the way in which the parties learned to compete within the presidential arena (see Bartolini, 1984). In Switzerland, the maintenance of the ‘magic formula’ has been facilitated by the depoliticization of government and the displacement of ultimate decision-making authority to the popular referendum. In the United States, the survival of a stable two-party system in the context of quite flexible and changeable political parties owes a great deal to the restrictive practices in electoral registration and ballot access. In the United Kingdom, two-partyism is helped by the combination of a plurality system of elections to the House of Commons and pronounced party discipline in Westminster. In other words, the institutions of politics provide us with the means and the language for thinking about political alternatives, and this also holds true for the party system, as well as for those institutions which work through the party system. They help to impart a language of politics which, when learned, is likely to become taken for granted and to resist change.

In addition, party system change will also be limited by the sheer stability of the wider institutional order within which it is nested. A party system, as Jepperson (1991: 151) notes of any given institution, ‘is less likely to be vulnerable to intervention if it is more embedded in a framework of [other] institutions’. And if these other institutions are themselves relatively stable, then it follows that the party system is more likely to remain intact. Conversely, a change in the institutional setting can provoke quite significant party system change. The case of Italy is the most obvious example here, in that the shift to the largely plurality voting formula in 1993 was one of the key factors that promoted the emergence of the new bipolar competition.

In sum, by focusing on the structure of competition for government, and by adopting a series of guidelines that indicate how that competition can change, we gain a perspective on party system change and stability that is not limited by the traditional categories that are found in the older literature, and that is more meaningful than the perspective offered by simple numerical summaries. There are four key advantages to this approach.

First, it enables us to make the crucial distinction between party system change, on the one hand, and party change, on the other, or between what Lipset and Rokkan (1967) famously referred to as the ‘freezing’ (and ‘unfreezing’) of party systems and the ‘freezing’ (and ‘unfreezing’) of individual parties. These are obviously two different processes, and can be applied quite independently of one another. As indicated above, parties may change quite substantially, and prove quite flexible and adaptable, and yet the party system can remain intact. Indeed, as I have suggested elsewhere (Mair, 1997: 16), party systems often manage to survive precisely because the parties refuse to be pinned down. At the same time, the parties can remain more or less the same, but because of a short-term change in elite preferences, or even because of a small change in voting patterns, their strategies can suddenly shift, and what had been a long-standing pattern of competition can suddenly be broken open. Britain, which has perhaps one of the most stable party systems in the world, came close to such a break in the lead-up to the 1997 election, when Labour considered a possible coalition with the Liberal Democrats (Webb, 2004: 26).

The second advantage of this approach is that it draws the emphasis away from the numerical criterion. When seen in very simplistic terms, party system change can be taken to be the same as numerical change—the system is different because the number of parties is different. This is very misleading, however, since, as argued above, numbers often fail to count in systemic terms, and they appear to count less and less as more and more parties join in electoral competition and win representation in parliament. In other words, while party systems have become more fragmented in recent years, this does not necessarily imply that the systems themselves have changed.

Third, and following from this, party system change conceived in these terms has the advantage of focusing on change at the core rather than at the margins. The stronger a party system is, the more easily we can see that it revolves around a core opposition—that it is ‘about’ something (Mair, 1997: 13-14). In Britain, the party system is about the conflict between Labour and Conservative; in France, it is about ‘left’ against ‘right’, often without a particular party specification; in Sweden, it is about the Social Democrats against the more loosely determined bourgeois bloc, and so on. In other words, the structure of competition, once established, becomes dominated by a particular choice, and other considerations become secondary. Conversely, to the extent that no single conflict manages to become established, the party system is likely to remain inchoate and ill defined. By tracing patterns in the competition for government, this distinction can easily be brought to light.

Finally, once we deal with party system change in these terms, it becomes relatively easy to pin down more or less precise moments in which prevailing patterns are suddenly broken—as was the case in Ireland in July 1989, in Italy in May 1994, in Germany in October 1998, and in Austria in February 2000. In other words, although it may take time before a new party system develops, and before a new structure of competition becomes established, a breakdown in the old pattern can be seen to have occurred quite suddenly and abruptly. And this, in turn, can open up the possibility of going on to treat party system change as an independent rather than just a dependent variable. It is one thing to identify the factors that lead to party system change, be these social, organizational or institutional, and allowing for the different conceptions of party system change that are used in the literature, this is a well-covered theme. It is quite another to trace the effects of party system change—whether on the component parties themselves, or on the voters—and this is something which, at least as yet, is a relatively neglected research question.

Party System Change: Future Trends

There are two important developments that have recently impacted on parties and on party systems and that are likely to shape the direction of party system change in the coming years. The first of these is what has been called ‘the victory of democracy’: the ending of the cold war, the collapse of the former Soviet Union, and the success of liberalism (or neo-liberalism) in establishing itself as a more or less universal source of governing principles. This is obviously a new and unprecedented situation, in which, as Perry Anderson (2000: 17) puts it, ‘there are no longer any significant oppositions—that is, systematic rival outlooks—within the thought-world of the West; and scarcely any on a world scale either’. The victory of democracy also makes for an enormous change in the context within which party systems operate, although its impact is sometimes difficult to appreciate. Given that the party systems in the long-established democracies grew up and became consolidated in an international context in which democracy was daily contested by non-democratic alternatives, and given that this international battle was often translated by competing political parties into a form of domestic opposition, the sudden ending of this conflict has had effects that reach into the heart of contemporary politics. One such effect has been the effective disappearance of any possible challenge to democracy at the domestic level—as Juan Linz (1997: 404) has noted, we are all democrats now, and ‘no anti-democratic ideology appeals to politicians, intellectuals, religious leaders … as an alternative to political democracy’—and hence the disappearance of the traditional anti-system party (see also Capoccia, 2002). With the demise or transformation of the former communist parties, and the effective disappearance of the traditional fascist alternative, there are no longer any important pariah parties in competition. In contemporary politics, in other words, and probably for the first time in democratic history, almost all parties have become salonfähig.

This obviously also now holds true even for the parties of the new radical right (see Bale, 2003; see also Mudde, 2001). These parties differ from the mainstream, sometimes dramatically so, in terms of policy, and often also in style, but they rarely differ in terms of any ultimate commitment to the maintenance of democratic procedures. As such, however unpalatable it sometimes seems, both domestically and internationally, mainstream parties do find it possible to forge compromises with these new parties and to bring them into government. Indeed, in recent years, this has become almost commonplace in Western Europe (see Heinisch, 2003), with the Freedom Party in Austria, the Lega Nord in Italy, and the Pim Fortuyn List in the Netherlands becoming full-fledged parties of government, and with recent centre-right minority governments in Denmark and Norway relying on support from Danish People’s Party and Progress Party respectively. The lesson here is evident: policy differences, however sharp, are always negotiable. The degree to which anti-systemness has now become a thing of the past was also made evident when the right-wing nationalist party Sinn Féin became part of the extraordinary coalition running the newly devolved government in Northern Ireland, even though its associated military wing, the Irish Republican Army, a terrorist organization, had not yet surrendered or decommissioned all of its weapons (Tonge, 2000).

The second important change that is likely to have an impact on the future direction of party system change has been the decline of parties as membership organizations, and the ascendancy of the party in public office within the party writ large (Katz and Mair, 1995, 2002). This general shift, which may be observed in traditional mainstream parties in almost all long-established democracies, and which, in its end state, is also often characteristic of emerging parties in both new and old democracies, has obviously many implications for how the individual parties behave and for how they communicate with and relate to the wider society (see Krouwel, Chapter 21, in this volume; see also van Biezen, 2003). This also has implications for party systems, however, the most important of which is that it helps to bring the parties closer to one another, This is especially true when judged from the perspective of the shared ambitions of comparable office-holding and office-seeking elites, and it is something that helps to push the parties further along the road towards becoming the sort of top-heavy campaigning organizations that are exemplars of the Schumpeterian or Downsian version of electoral politics: that is, teams of leaders who compete for the favour of the people’s vote (see Beyme, 1996; Farrell and Webb, 2000).

The combination of both of these factors suggests that the party systems of the future are more and more likely to reflect the type of bipolar competition that has long been characteristic of the French Fifth Republic, and that is now also clearly evident in Italy. This same pattern may also be emerging in Austria and Germany, and is already beginning to be established -whether in two-party or two-bloc form—in many of the newer democracies in southern and post-communist Europe, as well as in Latin America (Bale, 2003; Mainwaring and Scully, 1995b). This trajectory is likely for two related reasons. First, precisely because the substantive differences between mainstream parties are less pronounced, and because these parties can no longer function primarily as representatives in an ideological or purposive sense, elections will inevitably come to revolve more closely around the choice of persons rather than that of policies (see also Mair, 2003). Second, in order to help facilitate a choice of persons, and in order to ensure at least some degree of popular accountability and legitimacy, the parties will almost inevitably find themselves being driven towards bipolar competition, and hence towards an electoral process that affords voters a choice among alternative governments, and among alternative teams of leaders. Parties that govern have a clearer need for immediate electoral accountability than do parties that also represent, and unless accountability is promoted by the provision of clear alternatives in the electoral process, the parties themselves are likely to be seen as less legitimate. In this sense, we can see a version of the so-called Americanization of party systems—a downgrading of the role of policy in competition, and an enhancement of the role of personality, leading to the provision of clear choices between opposing candidates or teams of candidates. This is certainly the direction in which many party systems are now travelling.

It is therefore interesting that when we look across the contemporary advanced democracies, we see only a handful of polities that still appear to maintain a traditional pattern of overlapping coalitions with a more or less extended post-election negotiating period: Belgium, Luxembourg, the Netherlands and Finland. These countries, together with the special case of Switzerland, now constitute the only set of democracies where voters are effectively denied a direct say in the formation of the government. This is, in all likelihood, a dying tradition. Once parties present themselves to the public primarily as governors rather than as representatives, as is more and more often the case, then they will come under increasing pressure to cede to that public the choice of who will actually win office. In some cases, especially in the old consociational democracies, this may well require a major restructuring of the party system as a whole.

Two other changes may well go hand in hand with such a general shift in direction. First, a move towards bipolarity may well serve to facilitate party and party system renewal. If we assume continuing multipartyism, and hence also relatively permissive electoral systems, then bipolarity will almost certainly have to follow the pattern set by France or Italy rather than that of Britain or the USA. That is, it will involve competition between more or less flexible and malleable blocs of parties, or electoral alliances, rather than single parties or tight coalitions. Other things being equal, these blocs will be seen as being of the centre-left and centre-right, even though in practice there may be little in policy terms to choose between them. As in France and Italy, this will be what the party system is about—this will be the core. The result is that within each bloc it is likely that there will be less and less privileging of particular party organizations or labels, and more room for the free competition for influence and control. It is here that renewal will be found. The broad tendances will remain stable while the various parties reshuffle within them. Core coalitional continuity will coexist with an unsettled range of partisan or semi-partisan components.

Second, as elections increasingly come to revolve around personalities or teams of personalities, it is likely that the policy-making process will become ever more depoliticized. In other words, leaders who are chosen primarily because of their simple electoral appeal or celebrity status are more likely to seek to delegate decisions to non-political agencies. They are also more likely to be encouraged to do so by those around them. This process of delegation is already well advanced in most democratic polities, and was accelerated very substantially in the wake of the victory of democracy and the ending of the cold war. In practice, it means more decision-making power being ceded to so-called non-majoritarian institutions, a greater role for judges and other expert arbiters, and even more influence being accumulated by bodies such as the European Union, the World Trade Organization and the International Monetary Fund. Elections lead to office, but not necessarily to authority, and almost certainly not to expertise. As Schumpeter (1947: 288) once put it, ‘selection by means of success at the polls may work against people who would be successes at the head of affairs. And even if the products of this selection prove successes in office these successes may well be failures for the nation.’ Hence the perceived need for delegation.

In sum, the future trend for party systems may well be in the direction of a world that was already sketched by Schumpeter more than half a century ago: a world in which parties are teams of leaders, who compete for majority support in party systems that are increasingly bipolar, and who then rely on a host of expert agencies and international organizations to deal with their decision-making problems. Party systems may well become more competitive, and hence elections may well become decisive, but they also risk becoming less meaningful. Even if choice becomes more apparent, it may yet count for less.