Portal de Periódicos da CAPES

Is there a european electorate and what does it look like? evidence from electoral volatility measures, 1976–2004

2005; Taylor & Francis; Volume: 29; Issue: 1 Linguagem: Inglês

10.1080/01402380500389190

1743-9655

Daniele Caramani,

Electoral Systems and Political Participation

Is the EU party system a reflection of national electorates or a distinct arena based on specific alignments arising from the European integration process? The main indicators used to test for distinctive dimensions in European Parliament elections are indices of electoral volatility comparing national and European elections. Data include national elections over the last 30 years and European elections from 1979 until 2004 for all member states. Evidence shows persistently overlapping electoral behaviour due to the predominance in the two ‘orders’ of elections of the left–right dimension. The article argues that this similarity reveals a multi-level European party system. In historical comparison, it is shown that, despite different conditions of social and political mobilisation, the left–right alignment plays a similarly important integrating role in the ‘Europeanisation’ of electorates today as it played in earlier processes of ‘nationalisation’. Acknowledgements For insights and comments I am grateful to Ingrid van Biezen, Colin Hay, Liesbet Hooghe, Amie Kreppel, and Hermann Schmitt. Notes 1. See Van der Eijk and Franklin et al. (Citation1996) for a comprehensive analysis. See Hix (Citation2002) for an overview of the literature since the 1970s. 2. See Hix and Lord (Citation1997: 18). Kreppel speaks of a ‘transformation of the EP from a chamber of debate to a legislative body’ (Citation2002: 151). Hix, Kreppel, and Noury (Citation2003) come to the conclusion that the European party system is ‘ready for power’. 3. Often the level of mass electorates is neglected in integration studies, which focus instead on MEPs, party groups, and elite-level negotiations. For example, Hix, Kreppel, and Noury (Citation2003) distinguish an organisational and a competitive dimension in the EP, leaving aside the electoral process. 4. This comparison is, for example, mentioned in Schmitter (Citation2000: 70). For a more systematic juxtaposition see Raunio (Citation2002: 273). On ‘surge and decline’ see Campbell (Citation1960 and Citation1966) and Niemi and Weisberg (Citation1993: 207–21). 5. Explanatory factors accounting for the few cases in which turnout in European elections approaches national levels, include compulsory voting (Blumler and Fox Citation1982), positive views on Europe (Van der Eijk and Oppenhuis Citation1990), and time until the next national election (Marsh and Franklin Citation1996: 19). 6. Schmitt (Citation1990) has described the tendency of ‘wasting’ votes as ‘voting with the heart’. Others interpret voting cycles between first- and second-order elections in terms of a punishment effect against incumbents in national governments (Erikson Citation1988; Reif Citation1985a). On the ‘theory of second-order elections’ see also Reif (Citation1985b; Citation1997) and Marsh (Citation1998). Two more elements are presented in the literature on second-order elections which also emerge in Campbell's studies. First, the loss of votes for incumbent parties is caused by the lower turnout of the groups supporting incumbents in mid-term or European elections which on the contrary turn out for elections of incumbents’ offices. Second, second-order elections are interpreted as referenda on the performance of the incumbent government but not affecting its survival (Reif Citation1985b: 7–15). 7. Because of the focus on alignments, but also for reasons of space, turnout is not considered here. 8. The formula of the ‘Pedersen index’ (Citation1979) or ‘total volatility’ (Bartolini and Mair Citation1990) is the following: where P is the percentage of votes for parties ‘i’ to ‘n’ in elections ‘t’ and ‘t + 1’. Values of volatility are computed for each election. 9. Two national elections have been considered before the first European election because two (subsequent) elections are required to compute one volatility index. 10. In some cases national and European elections take place in the same year. If a previous national election was available, this has been preferred. In Luxembourg this is particularly problematic as the two types of election take place on the same day. Furthermore, the 2003 Dutch national election has been preferred as the 2002 vote was heavily influenced by the murder of a party leader. If the 2002 Dutch election had been taken, the solid curve in the Dutch graph in Figure 2 would have been flat between 1999 and 2002 instead of showing a decline since 1998. However, this choice has no impact on the argument presented here. 11. For data sources see notes to Table 2. Concerning national elections, in Germany the Zweitstimmen have been used (on the basis of which the allocation of seats is carried out by PR), and in Italy since 1994 the list votes have been used according to which seats are allocated proportionally (instead of the single-member constituency votes). For countries using two-ballot systems results for first ballots have been used. In Belgium, Walloon and Flemish parties have been added together for computational purposes. 12. NV data reproduce trends identified by previous research. Some minor differences are due to the choice of countries. Gallagher et al. (Citation2001: 260–65) include Norway and Switzerland (not considered here because they do not vote in elections to the EP). The inclusion of these two stable countries with traditionally low levels of volatility leads to a lower volatility also in Bartolini and Mair (Citation1990) up to 1984. In addition, these authors do not include Greece, Portugal, and Spain which are considered here since 1975 and were highly unstable until the early 1980s. Finally, data are not directly comparable with Dalton et al. (Citation2000: 38–42) because of their inclusion of non-European cases, as well as Norway and Switzerland. 13. In Portugal the NV was particularly high in the second half of the 1970s after the democratic transition and stabilises at the beginning of the 1990s (see graph in Figure 2). In the 1985 national election, it is mainly the rise of the Partido Renovador Democrático (with 18.5 per cent of the votes) that causes both high levels of NV (with respect to the 1980 election) and MV (with respect to the 1984 EP election). In the 1987 national election this party received 5.1 per cent of the votes nation-wide. 14. The anti-EC movement regroups candidates from different parties (although most of them are left-of-centre parties: Socialistisk Folkeparti, Socialdemokrater, Venstresocialisterne). The alliance includes the liberal Retsforbundet (Justice Party) as well. 15. The Radikale Venstre too scored less in European elections during the 1980s (they recovered strength in the 1990s) and the Centrum Demokraterne entered the electoral scene for the first time in the 1979 European election. 16. The Rassemblement pour la République (RPR) and the Union pour la Démocratie Française (UDF) were separate parties until they unified in 2002 as Union pour un Mouvement Populaire (UMP). The UDF still exists, formed by Liberals who did not join the UMP. 17. In part this was a consequence of the rise of the Centre pour l'Europe (Centre des Démocrates Sociaux) which received 13 per cent in the 1984 European election. 18. In the 1997 national election the Front National increased its support to 15.1 per cent. 19. A number of changes took place in the Irish party system – although they did not affect significantly the levels of volatility – principally with the rise of the Workers’ Party, the Democratic Left, and the Progressive Democrats. 20. In Germany the slightly higher levels of MV (3.27 more than NV) are caused mainly by the growth of the Greens in European elections and, in the late 1990s, by the bad performance of the Social Democrats. This would support the thesis of a ‘punishment’ of incumbent parties in second-order elections. The Sozialdemokratische Partei Deutschlands wins the 1998 and 2002 national elections but decreases to 30.7 per cent in the 1999 European election and to 21.5 in the 2004 European election. The Christlich-Demokratische Union receives less than 30 per cent in the 1998 and 2002 national elections and increases to 36–39 per cent in the 1999 EP elections. 21. For some cases the level of MV is lower than the average NV. Voters change preference more between national elections than between national and European elections (see also negative scores in the last column of Table 2). 22. The orthogonal dashed lines have been placed in the middle of the range between lowest and highest values of each of the two volatility distributions (extreme values concern Luxembourg and Estonia for the MV, and Malta and Hungary for the NV). 23. See Bartolini (Citation2002), Flora (Citation1999: 88–91), and Marks (Citation1997). Lodge (Citation1975) provides the first application of the ‘exit’ and ‘voice’ concepts to European elections. Most particularly in the legal sphere, the reduction of ‘selective exit’ through the jurisprudence of the European Court of Justice mainly, has led to the development of channels for the expression of voice (institutions of political representation) as well as to the development of political oppositions. See Weiler (Citation1999) for this interpretation of Hirschman's scheme. 24. As another functional dimension, Schmitter (Citation2000: 68) points to agriculture – a cleavage that has disappeared or has been incorporated in other alignments in national constellations – that at the EU level has re-emerged because of the resources for the Common Agricultural Policy and through alliances of ‘integration losers’ (economically weak groups and peripheral regions). Schmitter points to the salience of the territorial dimension at the EU level in the form of regionalist movements within and across national borders (Citation2000: 69). On the contrary, the religious dimension seems to have moved ‘outside’ Europe, with Christianity as a common element characterising European identity in opposition to other border cultures. 25. Such an approach might in particular prove useful in the explanation of the differences between member states in the degree of ‘differential’ of electoral behaviour. 26. These ideological schemes include not only revolutionary Socialism but also Nationalism associated with national independence and/or unification as well as democratisation (in the form of Liberalism as the third large ideology of the nineteenth century). 27. This runs against the argument (for example, Schmitter Citation2000: 66–71) suggesting that the European party system in the last two decades roughly resembles national politics in the mid-nineteenth century. 28. Methodologically, this is related to the assumption of the independence between cases. Rokkan's model itself is based on ‘deviations’ between countries in the development of party families. It is less concerned with the explanation of common aspects. Tilly's critique points precisely to Rokkan's failure in genuinely analysing the interactions between countries (Tilly Citation1984: 129; see also Flora Citation1999: 10). This is obviously different from – but not totally unrelated to – the so-called ‘n = 1’ problem (see Caporaso et al. Citation1997), that is, the explanation of unique phenomena such as ‘regional integration’ in Europe. 29. Within the vast methodological literature on ‘diffusion’ processes and on ‘Galton's problem’, see Klingman (Citation1980), Naroll (Citation1965), and Wellhofer (Citation1989).