How state-wide parties cope with the regionalist issue - FEB Leuven

Party competition in regional elections: The strategies of state-wide parties in Spain and the United Kingdom

Liselotte Libbrecht*, Bart Maddens* and Wilfried Swenden#

* Centre for Political Research, Catholic University of Leuven Parkstraat, 3000 Leuven, Belgium # Department of Politics, School of Social and Political Studies, University of Edinburgh Adam Ferguson Building, George Square, Edinburgh EH8 9LL, UK

Paper presented at the territorial Politics Conference of the Political Studies Association 10-11 January 2008, Edinburgh

Work in progress – Please do not cite

1

1. Introduction In a multi-levelled democracy with a strong regional tier of government, state-wide parties face an important electoral dilemma. On the one hand, electoral competition in state-wide elections requires that parties behave as single unitary actors and present a single and coherent programme to the electorate. However, in order to maximize votes in regional elections, statewide parties must give some leeway to their regional party branches in order to appeal to regional sensibilities. Reconciling state-wide with regional party interests is most problematic when the regional party systems differ substantially from the state-wide party system. One of the main causes of such highly differentiated dynamics is the substantial presence of autonomist parties. While these parties rarely play a big role in the state-wide political arena, they can weigh on the political debate in a devolved or federalised political system. For instance, in some Spanish regions they routinely collect up to or more than 50 percent of the regional parliamentary seats and as such increase the pressure on state-wide parties to react in one way or another. In this paper, we will discuss the strategic positioning of state-wide parties by focusing exclusively on their campaign content. To this purpose we have analyzed electoral programmes (‘party manifestos’) for two state-wide and sixteen regional elections in Spain, held respectively between 2000 and 2003, and between 1997 and 2003. Spain and the UK are both interesting case-studies since the regional assemblies are elected in separate elections and regional party-systems have developed which often contain one or more relevant autonomist parties (Pallarés et al., 1997). Rather than providing a general overview of the issue profile of state-wide parties in regional (and state-wide) elections, we analyze party manifestos to investigate how state-wide parties deal with ‘regionalist issues’ in state-wide and regional elections. Hence, the ‘regionalist issue’ profile of state-wide parties’ can be considered as our dependent variable. Regionalist issues are issues in which parties take a stance with regard to the present levels of regional autonomy or with respect to matters that touch upon regional identity and culture. When parties emphasize the former, we speak of ‘regional institutional issues’, where they debate the latter, we speak of ‘regional identity issues’. Again, Spain and the UK present excellent test-cases, given that some regions qualify as ‘minority nations’ or ‘historic communities’, thereby sparking a strong interest on regionalist issues in its institutional and identity dimensions. This paper is structured in four major parts. In the first part we derive a set of hypotheses from directional theories of issue voting and the comparative party organizational and federalism literatures to predict the issue profiles of state-wide parties in regional elections. The variables assumed to influence these profiles are the regional party system, the strength of the autonomist parties, the location of the various parties on the left/right-divide, the profile of state-wide parties in state-wide and other regional elections and the autonomy of the regional party branches. In the second part we reformulate these hypotheses with the UK and Spanish cases in mind. The third section provides a more detailed account of how we measure our dependent variable. In the final part we test the hypotheses on the UK and Spanish cases.

2

2. Theory and Hypotheses Defining the Dependent Variable Earlier we identified the ‘regionalist issue’ profile of statewide parties’ as our dependent variable. But how do we map the regionalist issue for a party. In this respect we rely on the insights of directional theory (Rabinowitz and Macdonald, 1989; Macdonald et. al., 1991).1 Research about voting behaviour and party competition used to be mainly concerned with the positional component. Traditional proximity theory assumes that parties compete by taking different positions on the same policy dimensions (see Downs, 1957; Enelow and Hinich, 1984). According to directional theory, by contrast, the strategic positioning of a party with regard to any issue involves two different components: a salience component and a directional component. The first component refers to the degree to which a party emphasizes the issue: it may keep silent about the issue and thereby dismiss it as unimportant, or it may stress the issue and thereby attempt to increase its salience amongst the electorate. The second component refers to the substantive content of the party’s issue profile, that is, the direction the party prefers with regard to the issue. Salience and Positional values combined make up Directional Intensity values. Therefore, whenever we refer to the overall regionalist issue profile of a party we refer to its directional intensity values (salience * position) on the two dimensions of the regionalist issue: the institutional and cultural components (a more detailed description of how the dependent variable is operationalized is provided in section 4). Adapting directional theory to the way in which parties deal with the regionalist issue, we can discern three main strategic approaches. In terms of salience, a party can choose to deemphasize the issue and attempt to prime other issues from which it hopes to benefit more. This is what Meguid (2004 and 2005) labels a dismissive strategy. We call this an ‘evasive’ strategy with regard to the regionalist issue. If, on the contrary, the party decides not to dismiss the issue, it has two options which both entail a positional element: either the party may take a regionalist direction (in terms of identity and/or autonomy), or a centralist direction. Applying Meguid’s terminology (2004 and 2005), in the former case it opts for an accommodative strategy in the latter case for an adversarial strategy. Both the accommodative and adversarial strategies will normally increase the salience of the regionalist issue in the election and thus its impact on the vote The crucial question of this paper is to explain the values on our dependent variable (and thereby implicitly also the chosen strategy of a state-wide party) when considering regional party competition in a multi-layered state. We take here a plausible scenario of two state-wide parties (most likely one left-wing, one right wing) which compete against each other. 1

More recent research has emphasized that parties compete not only by taking and adapting a position in the policy space, but also by attempting to manipulate the salience of the policy dimensions. Directional models of issue voting take into account both the policy direction which a party advocates and the intensity with which it does so. Furthermore, apart from making the salience component critical to assessing the role of issues, directional theory also rejects the spatial assumption that the policy stance of a party can be conceived as a specific position on a continuous policy scale. Instead, each issue is assumed to be essentially bi-directional. Issue ownership theory takes this reasoning a step further and dismisses the positional component altogether. From an issue ownership perspective, parties attempt to make the issues on which they are perceived to be most credible critical to the voter’s decision (Petrocik, 1990 and 1996). They differ by emphasising different issues rather than by taking different positions on the same issues. As a result of this lack of positional competition, all issues become valence issues in the sense that only one policy direction is emphasized (Budge and Farlie, 1983; Budge et al., 1987; Maddens, 1994).

3

Sometimes these state-wide parties will face the competition from one or several regionalist or autonomist party/parties. This gives rise to ten theoretically possible scenarios, as summarised in Table 2. The importance of the regionalist issue in the campaign ranges from zero (when the state-wide parties do not face a regionalist competitor and both opt for an evasive strategy) to 3 (when a regionalist competitor is present and both state-wide parties opt for either an adversarial or an accommodative strategy). In four of the ten scenarios, whenever at least two parties emphasize opposite directions, the issue functions as a position issue. If only the regionalist direction is stressed by at least one party (which is the case in five of the ten scenarios), it becomes a valence issue. A scenario involving a monopoly of the centralist direction – and a reversal of the valence - is considered implausible, because a state-wide party adopting an intense centralist stance would probably pave the way for a regionalist party. Table 2 Hypothetical scenarios concerning the strategic positioning of two state-wide parties in regional elections with regard to the regionalist issue Regionalist StateState-wide Status of Importance Party wide Party B regionalist of Party A issue regionalist issue (0-3) 1 REG REG REG Valence 3 2 REG CEN CEN Position 3 3 REG CEN REG Position 3 4 REG EVAS REG Valence 2 5 REG EVAS CEN Position 2 6 REG EVAS EVAS Valence 1 7 REG REG Valence 2 8 CEN REG Position 2 9 REG EVAS Valence 1 10 EVAS EVAS Non-issue 0 REG = emphasis on regionalist direction, CEN = emphasis on centralist direction, EVAS= no emphasis on regionalist issue.

We can hypothesize that the values of our dependent variable will be affected by the following independent variables. (1) The nature of the regional party system - A regional party system without autonomist parties The strategy of state-wide parties will depend on the nature of the regional party system and particularly the strength of so-called regionalist parties therein. Where state-wide parties do not face the competition of one or several sizeable regionalist parties (scenario’s 7-10), one possibility is that they will try to keep the regionalist genie in the bottle and avoid the regionalist issue, as a result of which it simply will not materialize in the campaign (scenario 10). But an alternative strategy is to pre-empt the future rise of regionalist parties by occupying the regionalist niche in the electoral market and by emphasising a regionalist stand, which could be defined as an anticipatory accommodative strategy. The allocation of issue ownership is assumed to take place on a ‘first come first serve’ basis, in the sense that the party which is the first to adopt a credible profile on the issue becomes the owner (Petrocik, 1990: 5-6), leaving the other parties with no other option but to downplay the issue (scenario

4

9) or respond with an adversarial, that is, centralist strategy (scenario 8). Finally, as argued above, it seems unlikely that a state-wide party would do the reverse and oppose an anticipated regionalist party by stressing a centralist stand. Such a strategy would probably function as a self-fulfilling prophecy and give rise to a regionalist party. - A regional party system with autonomist parties Where state-wide parties are faced with an autonomist competitor (scenario’s 1-6), directional theory contends that the regionalist vote will be captured by the party that takes the most intense regionalist stand, provided it remains within a region of acceptability, that is, is perceived as responsible by the voters. This implies that a vote maximising state-wide party faced with a significant regionalist competitor has little to gain from an accommodative strategy that stops short of putting as much emphasis on the issue as the regionalist party2. If the state-wide party is not prepared to do so –and we can assume this to be the case, given that such a strategy would upset the state-wide party - , it is left with the option of either downplaying the issue or of playing the centralist card. The latter option, however, is probably unlikely in a regional election. Due to the inherently regional focus of such an election, we may expect that there will be a certain regionalist atmosphere in the campaign which may cause public opinion to tilt towards the regionalist direction. As a result, it may be all the more hazardous for a state-wide party to adopt an adversarial centralist stand, thereby swimming against the tide. According to this reasoning, the only choice left is to opt for an evasive strategy and to divert voters’ attention to other issues on which the party is perceived to be more credible. This is what we would expect to happen on the basis of issue ownership theory: the owner – for instance the regionalist party – attempts to increase the salience of the issue, while the other parties attempt the reverse (scenario 6). In sum, we anticipate that the dominant strategy for state-wide parties – irrespective of the size of the regionalist challenger – will be an evasive one, i.e. the state-wide party will give the regionalist issue a significantly lower salience than the regionalist competitor and this salience will not vary with the size of the regionalist party. Hence we can hypothesize the following: H (1) Irrespective of the size of an autonomist party; in a context where state-wide parties compete against one or several autonomist parties, the most likely strategy is one in which the state-wide parties evade the regionalist issue. (2) The location of the autonomist party on the Left-Right Axis On the basis of the arguments discussed above, it seems rather unlikely that a vote maximising state-wide party would opt for an adversarial centralist strategy, irrespective of whether there is a regionalist competitor in the field. Yet, applying Meguid’s (2004 and 2005) reasoning, it could be argued that a state-wide party may still opt for such a strategy in a setting with a regionalist party, notwithstanding that public opinion is expected to lean to the 2

In contrast to directional and issue ownership theory, however, Meguid (2005: 349) argues that an accommodative strategy may challenge the exclusivity of a niche party and undermine its issue ownership and thus its electoral support. Her empirical findings (regarding the environmentalist and radical right parties) indicate that an accommodative strategy has a weak negative effect on the electoral success of the niche party, while an adversarial strategy has a much stronger positive effect. This model is based on a categorical distinction between the three strategies and does not take the intensity of an accommodative or adversarial stance into account.

5

regionalist side in a regional election. The aim of the strategising party would then be not to maximize its own vote, but to harm the other state-wide party by increasing the salience of the regionalist issue and thereby giving an advantage to the regionalist party. Such a scenario (scenario 5) could materialize if there are reasons to believe that the targeted state-wide party is more vulnerable to the regionalist party than the strategising one. That could for instance be the case if the targeted party is closer to the regionalist party with regard to other issues, as indicated by its position on the left/right spectrum. On this basis, we can hypothesize the following: H (2) Notwithstanding H1, when one of the state-wide parties is close to the regionalist party on the left/right divide, a state-wide party located either at the other side of the divide or in the centre might be expected to emphasize the regionalist issue not in order to win votes but to cause the targeted state-wide party to loose votes to the regionalist party. (3) The strategy of the state-wide party branch Clearly, the strategies adopted by the state-wide parties in the regional arena will also depend on the broader political setting, more specifically, their compatibility with the strategies of the state-wide party branch or with other regionalist branches on the regionalist issue. To take an extreme example, a state-wide party would lose credibility if it were to adopt a salient and pro-centralist position in state-wide elections and a salient and highly pro-regionalist position in regional elections. A state-wide party that takes the former position in state-wide elections can attempt to downplay the issue in regional elections, but realistically it cannot be expected to play the regionalist card. More in general, the leeway a party has in regional elections can be hypothesized to depend on the profile adopted on the issue in state-wide elections. Therefore, H (3) The more intense the party defends either direction in state-wide elections, the smaller the odds that a party can take an opposite stand in regional elections. (4) The autonomy of the regional party branches While the strategy of the state-wide party branch is likely to cast a shadow on the strategies of the regional branches, the leeway of regional party branches to ‘divert’ from the state-wide party with respect to the regional issue is likely to be affected by their level of autonomy. Indeed, manifestos for state elections are normally drafted by central party bodies, but there may be important differences as to which body drafts regional party manifestos. It may be the regional party branches alone, the central bodies of the party or both. In theory, for electoral reasons, a central party branch may want its regional campaigns to divert from the state-wide election campaign. However, we hypothesize that in general: H(4) An organizationally more centralized party is less likely to tolerate divergence between state-wide and regional manifestos (on the regionalist – and other – issues) than an internally decentralized one.

6

3. Case Selection and Case-Specific Hypotheses Case Selection We select Spain and the UK for an in depth analysis of the hypotheses that were listed above. These are both multi-national and multi-layered states with directly elected regional parliaments/assemblies. In Spain, regional elections are held across all seventeen autonomous communities. In the UK, devolved elections have been held since 1999 in Scotland and Wales (although devolution to Northern Ireland has been restored after elections in March 2007, British state-wide parties are not represented in Northern Ireland, as a result of which it was excluded from the analysis). However, for practical reasons we limited the analysis in time and, in the Spanish case, also in the number of selected regions. For instance, for Spain we considered (state-wide and regional) election manifestos in the period 2000-2003 for the two largest state-wide parties, the PSOE and the PP. For the UK we considered state-wide and regional elections between 1997 and 2003, for the three largest state-wide parties: the Conservatives, Liberals and Labour Party. In Spain, a selection of eight Spanish regions was made with the aim of maximising the variance on some of the key explanatory variables that were identified above: the nature of the regional party system (and especially the strength of regionalist parties therein) and the relative position of the state-wide parties on the left/right-divide vis-à-vis the regionalist parties. For the United Kingdom, we can offer a more ‘complete’ case-selection, since we include two regions with an elected regional chamber and state-wide party presence, i.e. Scotland and Wales, and therefore, only exclude manifestos for elections of the Greater London Assembly (for which only two elections have been held thus far). The selection of eight out of seventeen autonomous regions is primarily based on the relative strength of the regionalist party or parties therein. Table 2 (adapted from Pallarés, 1994) includes both the percentage which the regionalist parties obtained in all regional elections preceding the investigated period (2000-2003) as the average percentage for these elections. A first group contains three regions with strong regionalist parties, obtaining at least about half of the votes (listed as Type I regions in Table 2) Five regions have medium-sized regionalist parties, with an average electoral support from just below 20% to 33% (Type II regions in Table 2) Six regions have significant but small regionalist parties which poll generally less than 10% of the vote (Type III regions in Table 2). In the three remaining regions, there is no significant regionalist party. (Type IV regions in Table 2) In most regions, the regional party system is similar to the federal (quasi-) two-party system: the PP and the PSOE are dominant and compete for an absolute majority of seats. The stronger the regionalist party, the larger the odds that neither of the dominant parties gets a majority and the regionalist party obtains a pivotal position. Four regional party systems deviate from this pattern as one of the two leading parties is a regionalist party. This is the case in Catalonia, the Canary Islands, the Basque Country and Navarre. Galicia is a borderline case as, up to the 2003 election, the regionalist BNG and PSOE always competed for the second place, with the PP as clear frontrunner (Ocanã and Oñate, 2000). Table 2 About Here The eight regions that were selected for comparative analysis were drawn from each type or group of regions. For instance from group I, we selected the Basque Country and Catalonia. From group II we selected the Canary Islands and Cantabria. From Group III we selected La Rioja, Asturias and Castile and Leon. We did not for instance select the important region of Andalusia from this group, since we decided to limit the analysis to regions with

7

regional elections that do not concur with state-wide elections. Finally, we selected Murcia from among the small cluster of Group IV regions. Given its atypical status as state capital, we decided not to select the autonomous community of Madrid from within this group. Furthermore, Castile-La Mancha had to be excluded as a possible case because the regional PSOE branch PSOE did not have a manifesto for the 2003 election. Using the same criteria, we can establish where Scotland and Wales would fit. Table 3 clearly shows that both regions would be Type II regions, since the leading autonomist parties (Scottish National Party in Scotland and Plaid Cymru in Wales) collected roughly between 20 and 30 percent of the vote in regional elections. Because of the strength of the autonomist parties in the Scottish or Welsh party systems, the state-wide parties are less well represented there than in the state-wide parliament. Tables 3 About Here

8

Table 2 Support for regionalist parties in regional elections in all Spanish autonomous communities Type of Region

Region

1983

I

Basque Country (1980-1984-1986-1990-1994-1998) †

1980 64.47 43.38 1980 39.35 16.84 20.64 6.67 12.98 22.18 3.09 1982 5.41 7.52 8.52 2.82 2.85 -

Navarre Catalonia (1980-1984-1988-1992-1995-1999) II

III

IV

Canary Islands Aragon Cantabria Galicia (1981-1985-1989-1993-1997) Balearic Islands Valencia Andalucía (1982-1986-1990-1994-1996-2000) La Rioja Extremadura Asturias Castile and Leon† Murcia Castilla-La Mancha Madrid

1984 64.64 1984 51.47

1986 6.78

1987

1991

1995

1999

Average support for regionalist parties*

67.90 52.08

67.39 58.07

59.17 67.39

55.87 70.93

63.03 58.37

50.18 31.91 29.54 13.00 22.97 15.41 9.24

54.79 37.26 27.33 40.57 16.82 11.34 14.18

50.93 37.87 25.69 32.8 18.55 17.6 9.82

46.81 42.38 24.82 17 25.11 19.39 9.36

48.92 33.25 25.60 22.01 19.29 17.18 9.14

10.80 6.50 5.86 1.79 3.03 3.82 -

5.85 5.47 4.02 2.77 3.07 3.06 -

6.7 6.82 3.86 3.22 3.23 -

7.53 5.90 2.85 9.88 5.23 -

7.18 6.44 5.02 3.53 3.48 1.95 -

Source: compilation of data from the Archivo Histórico Electoral, Presidencia de la Generalitat Valenciana: http://www.pre.gva.es/argos/archivo/index.html *Source: compilation of data from the Archivo Histórico Electoral, Presidencia de la Generalitat Valenciana: average of all regional elections 1980-1999, this means 6 regional elections for the Basque Country, Catalonia and Andalusia and 5 for all the other regions. Only if the sum of support for all regionalist parties in one region is ≥1%, this is taken into account in the calculation of the region’s regionalist party support. # These percentages include the votes for UPN (Unión del Pueblo Navarro), which is closely allied with PP. Both parties have an agreement according to which only UPN competes in regional elections and only PP in the state election. It is a matter of dispute whether UPN should be considered a separate ‘regionalist’ party. If the party is not included, the percentages change to 28,6% (1999) and 26,9 (average), as a result of which Navarre shifts to the second category. † The ‘regionalist’ parties in Castile y Leon are in fact ‘sub-regionalist’ parties in the sense that there is a party defending the interests of Castile (Tierra Comunera-Partido Nacionalista Castellano – TC-PNC; 1.42% in 1999) and another one defending the interests of Leon (Unión del Pueblo Leonés – PNC; 3.81% in 1999). When these parties are considered sub-regionalist and not regionalist, Castile and Leon drops to group IV. The Unidad Alavesa in the Basque Country defends the interests of the people from Alavés (this party was dissolved in 2005) and is also a ‘sub-regionalist’ party. In order to be consistent this party will also be included in the calculation of the percentage of regionalist votes.

9

Table 3. Support for autonomist parties in regional elections in Scotland and Wales, 19992003 Type of Region

Region

1999

2003

II

Scotland

27.3

20.9

II

Wales

30.5

19.7

Source: House of Commons Library Research Paper 03/59 (2003) UK Election Statistics: 1945-2003 http://www.parliament.uk/commons/lib/research/rp2003/rp03-059.pdf

Case Specific Hypotheses: We can now adapt some of the hypotheses that were listed above with respect to case specific expectations from Spain and the UK.

(1) The location of the regionalist party on the Left-Right Axis Although the number of regionalist parties in the UK is limited the opposite is true for Spain. For starters, some Spanish regionalist parties are clearly ‘nationalist’ (such as the PNV, CiU, BNG…) in the sense that they proclaim a national status for their autonomous community and consequently have an electorate tending towards an exclusively regionalist identity and an ethno-nationalist attitude. These parties are located on the ‘ethno-nationalist’ side of a territorial cleavage, forming a separate ideological dimension in the space of electoral competition. Other regionalist parties are regionalist in character simply because they exclusively campaign within the region, but they are likely to take a much more moderate stance on extending levels of regional autonomy or appealing to regional identity issues, in tune with the electorate of their region. In these cases, the rise of regionalist parties has resulted from the political opportunity structures created by the new regional institutions and by the particularities of the regional party system. The latter are often related to the fact that the disappearance of the UDC and the CDS created a vacuum to the right of the centre, which could be filled by a regionalist party (Pallarés et al., 1997: 167-8). Second, the eight Spanish regions represent a broad variety of patterns of competition between regionalist and state-wide parties (Table 4). Three of the selected regions have regionalist parties that are located on the centre right of the political space and compete mainly against the PP. This is most clearly the case in Asturias (Buznego, 1998; Martín, 2003) and Cantabria (Ramos Rollón, 1998; Molina, 2003), where conservative regionalist parties were formed as a result of a split within the PP, respectively, the Uníon Renovadora Asturiana (URAS) and the Uníon para el Progreso de Cantabria (UPCA). An older regionalist party in Cantabria, the Partido Regionalista de Cantabria (PRC), is also generally perceived to be equally conservative as the PP, partly because it has preferred to ally with the PP rather than with the PSOE (Ramos Rollón, 1998: 167-8).

10

Table 4: Position of regionalist parties in eight selected regions and percentage of vote for regionalist and state-wide parties in previous regional election Basque Country (Type I) Catalonia (Type I)

Left/Centre left EA (8.7%) HB (17.9%) PSE-EE/PSOE (17.6%) ERC (8.8%)

Centrist

PP (20.1%) CiU (38%)

PSC-CIPC (38.2%)

PP (9.6%)

Canary Islands (Type II)

CC (37.5%) FNC (4.9%) PSOE (24.4%)

PP (27.6%) PRC (13.8%) UPCA (3.1%) PP (43.6%)

Cantabria (Type II) PSOE-Progresistas (33.9%) La Rioja (Type III) Asturias (Type III)

Right/centre-right PNV (28%)

PR (6.4%) PSOE (36,1%) PAS (2.6%)

PP (52,4%) URAS (7.3%)

PSOE (46.7%) Castile and Leon (Type III)* PSOE (33.9%) Murcia (Type IV) PSOE (36.%) Bold: regionalist parties * cf. note on Castile and Leon in table 2

PP (32.8%) PP (52%) PP (53.6%)

Other regionalist parties are to be characterized as catch-all parties drawing votes from either side of the political spectrum. The best example is probably the Coalición Canaria (CC). Until 1993, the Canary Islands had a variety of constantly changing regionalist, in many cases even ‘insular’, parties. This fragmentation of the political landscape came to an end in 1993, when six of the regionalist parties formed the CC. The coalition contains both left-wing and more conservative parties who had managed to capture the sizeable UCD-legacy in the eighties (Hernández-Bravo, 1998; Soriano, 2003). At the end of the nineties, an alternative and much smaller regionalist coalition was formed in the Canary Islands: the Federación Nacionalista Canaria (FNC). This party can also be considered as centrist, as it contains a broad array of mainly insular parties who had either kept out of or split off from the CC. Convergència y Unió (CiU) in Catalonia occupies a comparable position as the CC in the Canary Islands. The CiU also originated as a coalition of parties, ranging from the centre-left (Convergència Democràtica de Cataluña – CDC) to the centre right (the Christiandemocratic Unió Democràtica de Cataluña – UDC) and has developed as a broad catch-all party, drawing votes from both the centre-right and the centre-left of the political spectrum (Soler, 1998; Díez-Medrano, 1995). The Christian-democratic Partido Nacionalista Vasco PNV occupies a somewhat similar position in the Basque Country. However, when in 1986 its left-wing split off and formed Eusko Alkartasuna (EA), the PNV lost some of its centrist catch-all character. However, during the nineties both parties gradually grew closer and in the 2001 regional election they formed an alliance (PNV-EA). At the same time, the ethnonationalist conflict is so entrenched in the Basque Country that it is extremely difficult to draw a parallel with other regions. A much smaller centrist party is the Partido Riojano (PR). The party was founded as the Partido Riojano Progresista but quickly adopted a centrist profile as indicated by its attempt, in 1987, to form an alliance with the UCD (Fernández Ferrero, 1997: 104). In spite

11

of its modest electoral score, the party plays an important pivotal role in the political system, first pacting with PP (from 1987 to 1990) and then shifting its allegiance to PSOE (from 1990 to 1995) (Llamazares and Reinares, 1998 : 319-20; Mora, 2003 : 324-5). Finally, other regionalist parties have to be located on the left side of the political spectrum and mainly compete with the PSOE. Esquerra Republicana de Cataluña (ERC) (Soler, 1998) and Partido Asturianista (PAS) (Martín, 2003) belong to this category. The Basque Eusko Alkartasuna (EA) – whose ideology is sometimes characterized as socialliberal - and Herri Batasuna (HB, now banned) can also be considered as (respectively centre and radical) left-wing parties (Llera, 1998), but in this case too the same cautionary note applies as mentioned above (unlike for the other regions the dominant dimension of Basque party competition centres on the regionalist and not the left-right issues). Compared with the Spanish regionalist parties, the British regionalist parties are much easier to categorize. Both the Scottish National Party and Plaid Cymru are left-wing and nationalist parties. Since they share this orientation with the Labour and Liberal parties, the Welsh and Scottish party systems are more left-wing than the UK state-wide party system. Table 5 summarizes their position. Table 5. Position of regionalist parties in Scotland and Wales* Left/Centre left Centrist Right/centre-right Labour Conservative Party Lib-Dem SNP Wales Labour Conservative Party (Type II) Lib-Dem PC *Unlike in the Spanish case we can not add the election results for previous regional elections since the ones in 1999 were the fist ones Scotland (Type II)

Given what we know about the nature of the regional party systems and the ideological position of the regionalist parties on the left-right divide, what could we expect in terms of the strategic positioning of both state-wide parties on the regionalist issue? In the regions where both the PP and the PSOE are in competition with one or more substantial and centralist regionalist parties (Canary Islands and Catalonia, but not the Basque Country where the votes do not float between the PP and the regionalist centre party), we expect the dominant ‘evasive’ strategy to prevail. However, in regions where the regionalist/state-wide competition is located either on the left or the right of the political space, directional theory (complemented with Meguid’s insights) would predict that the opposite state-wide party will be tempted to draw voters away from its state-wide opponent to the regionalist party by emphasising the regionalist issue (either in a regionalist or a centralist direction) and thereby making it more central to the voter’s choice (cf Hypothesis 2). Cantabria would be the most obvious battleground for the PSOE to attempt such a strategy. The PSOE would be expected to prime the issue so as to draw votes away from PP to the right-wing regionalist party. An accommodative strategy would be most in line with the PSOE’s generally favourable stance towards decentralization (cf.infra). Asturias is another region where a strategy aimed at prioritising the regionalist issue could pay off for the PSOE, as the tiny PAS hardly has an impact upon the dynamics of party competition and does not pose a big threat to the socialists. The most likely region for the PP to attempt such a strategy aimed at harming its state-wide opponent by priming the regionalist issue is the Basque Country. Because the ethno-nationalist cleavage is so entrenched in this region, votes do not float between the centrist PNV and PP. On the left side of the spectrum 12

there is more fluidity as the PSE-EE (the Basque PSOE, which in 1993 merged with the leftwing nationalist party Euskadiko Ezkerra – EE) has stronger links with Basque nationalism (Llera, 1998). As a result, it could be in the interest of the PP to emphasize the regionalist issue in order to make PSOE loose votes to the regionalists. In that case, an adversarial strategy (taking a strong centralist position) would be the most likely option given the PP’s more centralist ideology. A similar observation can be made with regard to British state-wide parties. The autonomist parties pose a stronger challenge to Labour and Liberal Democrats than to the Conservatives because they are ideologically closer to Labour and the Liberal Democrats on the left right scale (Webb 2000: 19). Therefore, the Conservatives (as the only right-wing party) may wish to harm their largest state-wide competitors by opting for an adversarial strategy on the regionalist issue instead. This way the Conservatives may wish to raise the importance of the regionalist issue in the campaign, and by doing so increase the salience of the regional-autonomist issue to divide the left-wing camp. (2) The Influence of State-wide Party Ideology As discussed above, we also expect the profiles of the regional party-branches to be constrained by the strategy of the state-wide parties. For the Spanish case, we know that the PP due to its Francoist roots and its conservative ideology has traditionally been the main defender of Spanish unity, while the PSOE has always been somewhat more accommodative towards the regionalist demands. Therefore, we expect that the stands of the PSOE and the PP in state-wide elections will slightly tend towards the regionalist and the centralist direction respectively. As a result, the leeway of the regional branches to adopt an intense profile in the other direction should normally be limited. Similarly, for the British case, we know that after its initial rejection of devolution, the Conservative party has accepted the institutional status quo. The state-wide party has problematized the ‘West Lothian question’ and defends the existence of the union. The statewide branch of the Labour party also defends the institutional status quo and sees devolution as a way to save the union. The Liberal Democrats consider that devolution is a first step in the direction of a federal UK. As a result, we expect that the stands of the Labour Party will tend towards the status quo or at best slightly regionalist direction, whereas that of the Liberal Party will clearly be more regionalist. Finally, the Conservatives are likely to occupy a position that corresponds with the status quo or slightly tends to a centralist direction. In light thereof, the leeway of regional party branches to adopt an intense autonomist profile (in the case of Labour and the Conservatives) is expected to be limited. (3) The Autonomy of the Regional Party Branches The two main state-wide parties in Spain, PP and PSOE are both fairly centralized parties, even though the PP insists more on party cohesion and national uniformity than the PSOE. Thus, we may expect more similarity between the issue profiles of its manifestos across the various regional elections and between the state-wide and regional elections in the PP than in the PSOE (Pallarés and Keating, 2003). The Catalan PSC occupies a special (semiautonomous) position within the PSOE and can freely draft its manifesto for regional elections. Therefore, the difference between state-wide and regional manifestos with regard to the regional issue should be largest for the PSOE in Catalonia, and smaller for the other regions. In the UK, a slightly different picture emerges from what is found in Spain. The Labour party is the party with the least autonomous regional branches. While the regional

13

branches of Labour have some autonomy over policy-making for regional elections, the central party, in particular via the Secretaries of State for Scotland and for Wales, can influence the making of regional party policy (Fabre et al., 2006). In contrast with Labour, the Liberal Democrats define themselves as a federal party with regional branches that are free to select their own political personnel and decide on the content of their manifestos. As a result of this federal conception of party organisation, the central party does not intervene in the process of regional policy-making. This freedom is not, of course, completely unchecked, in the sense that the regional branches try to retain a degree of coherence with the state-wide party, so that their policies may be 'different but not dissimilar' (Fabre and Maddens, 2007). Finally, the Conservative party is positioned somewhat between Labour and the Liberal Democrats in terms of the autonomy of their regional branches. The regional elites, however, are imbued with a strong sense of political loyalty toward the UK Conservatives. Overall, the regional branches follow the central party line for non-devolved matters but have been given considerable free reign to develop their own policies over devolved areas (Seawright, 2004, Fabre and Maddens, 2007). On these grounds we would expect the regional profiles of the Liberal Democrats to be the most distinctive and that of the Labour Party to be the least distinctive. The Conservatives are expected to occupy an intermediary position. 4. Operationalizing the Dependent Variable: a more detailed description. In order to map ‘regionalist issues’ we consider the profile of parties on the basis of their manifestos3. The text of the manifesto was first split up into separate statements or quasisentences, applying the Comparative Manifesto Project (CMP)-guidelines (Volkens, 1992; Budge et al., 2001). Next, the statements pertaining to the regionalist issue are counted. For the purpose of this research we did not apply the CMP coding scheme, but simply identified the statements related to the regionalist issue4. The latter were defined as statements that in one way or another deal with the relationship between the regions and the centre, either because they involve the division of competences and/or the balance of power between them (the institutional component), or because they involve policies relating to strengthening the culture, language and identity of either of them (that is, the cultural component). Following directional theory, we cannot limit the analysis to the salience component (as in the CMP approach), but need to ‘unpack’ their campaign messages according to salience, position and a combination of both (directional intensity). The salience of the regionalist issue can be operationalized in a straightforward way as the share of statements or ‘quasi-sentences’ on regional autonomy and identity of all statements or ‘quasi-sentences’ in the manifesto. The substantive position of a party with regard to the regionalist issue requires a more sophisticated procedure involving a qualitative assessment of statements according to the 3

Manifestos are a primary data source: most electoral debates in secondary sources, such as the mass media, would concentrate on policy pledges contained in the manifestos (Ashworth 1999). Budge et al. (2001) refer to political texts as ‘the major source of evidence … for how democracy functions’. Moreover, manifestos are major statements of parties intended to present the full picture of their stances; they are issued by the whole party and not by one particular faction. Ashworth (1999) adds that party manifestos are publicly available and easily accessible, and they can be easily re-analysed. Moreover, in contrast with interview data, party programmes are well suitable for quantitative content analysis. 4

Apart from this broad distinction between regionalist and non regionalist statements we also applied a more fine-grained coding scheme, based on the Eurovoc thesaurus, which reflects the specific policy fields allocated to the competence of either the regional or national government. These data are not used in this paper.

14

directional cues which they provide. Following directional theory (Rabinowitz and Macdonald, 1989; Macdonald and Rabinowitz, 1993: 65), we assume that a party’s position can be defined as the probability that a party will prefer one direction (for instance, more competences for the region) over the other (such as more competences for the centre). This probability or ‘directional certainty’ ranges from –1 (certainty that the party prefers one direction) over 0 (probability of preferring one option equals the probability of preferring the other) to 1 (certainty that the party prefers the other direction). For the purpose of measuring this directional certainty of the regionalist issue, a coding scheme was developed distinguishing between five main categories: (1) centralist, (2) status quo centralist, (3) status quo, (4) status quo regionalist, (5) regionalist. As concerns the institutional dimension, statements expressing a preference for more competences for the region (category 5) can be considered as clear cues that the party favours the autonomist direction, while the reverse can be said of statements favouring a recentralization of certain competences (category 1). With regard to identity matters, category 5 and 1 imply that the party unequivocally proposes to strengthen and promote respectively the regional and the national identity. Category 3 contains neutral or highly ambiguous statements that provide no cue whatsoever as to the preferred direction and can be considered as an implicit stance in favour of the status quo, either with regard to institutional or identity matters. Taking only these categories into account, the directional certainty variable can be measured as the difference between the number of clear directional statements in one sense minus the number of clear directional statements in the other sense, divided by the total number of relevant statements. In this way the certainty would equal (minus) one if all the relevant statements provided consistent directional cues, or zero if all relevant statements fell in the neutral category or if statements in one sense cancelled out statements in the opposite sense. From the perspective of directional theory the interpretation of the categories 2 and 4 is less straightforward. These statements do not explicitly express a preference for a change of the status quo in either sense, but they contain a positive assessment of the status quo, from a more regionalist (category 4) or a more centralist (category 2) perspective. The status quo is advocated because it is considered to strengthen the identity or the institutions of the region (category 4) or the centre (category 2). While the voter obtains no clear cue as to the preferred policy direction, it can still be argued that he or she may ‘hint’ which side the party is more sympathetic to. This could be taken into account by weighing these statements by half. Directional certainty could thus be measured as the number of clear directional statements in one sense plus half the number of hints in the same sense minus the number of clear directional statements in the other sense minus half the number of hints in the other sense (the directional balance), divided by the total number of relevant statements. Directional Certainty = ∑(Cd1 + (Hd1/2)) - (Cd2 + (Hd2/2)) ∑ Sr with Sr the number of relevant statements, Cd1 the number of clear directional statements in one direction, Hd1 the number of hints in the same direction, Cd2 the number of clear directional statements in the other direction, and Hd2 the number of hints in that same direction.

15

Finally, directional theory asserts that in order to fully assess a party’s strategy on an issue we need to take into consideration their salience and position with regard to that issue. Bringing salience and position together will allow us to draw a measure of directional intensity. For instance, if a party unequivocally indicates that it prefers the direction of more autonomy for the region, but hardly emphasizes this stance (in the extreme case, it may just mention it once) its impact on voting behaviour will be very low. Conversely, the same observation applies to a party which allocates a lot of emphasis on the issue, but conveys highly ambiguous or contradictory messages. Both scenarios result in a low intensity issue profile, as a result of which the issue will not have a (positive or negative) impact on the evaluation of the party by the electorate. Conversely, an issue will only affect the vote to the extent that a party allocates sufficient emphasis on the issue and spreads clear directional cues. The intensity of a party’s profile can be measured as the product of directional certainty and the salience of an issue (Macdonald and Rabinowitz, 1993: 65; Rabinowitz and Macdonald, 1989: 118). As this value equals the directional balance divided by the total number of statements in the manifesto (times 100), it is not affected by the number of neutral or ambiguous statements. In other words, a party may devote its entire manifesto to the regionalist issue (and thus arrive at a salience value of 100%), but if there is not even a hint as to the preferred policy direction or if these hints are contradictory, the intensity will remain zero. Directional intensity could range from -100 to +100. A party which devotes its entire manifesto to the regionalist issue and favours more autonomy for the region in every statement would obtain a score of 100. 5. Analysis We can now test the higher formulated hypotheses with the UK and Spanish cases in mind. We do so by ‘unpacking’ the dependent variable in its two component parts (salience and position) and by considering institutional and identity values on the regionalist issue separately. Salience of the regionalist issue in Spain and the UK Table 6. Salience of the regionalist issue: Spain Type of region

PP Salience

State election 2000 I II III

IV

Basque Country Catalonia Canary Islands Cantabria La Rioja Asturias Castile and Leon Murcia Mean regions

PSOE

institutional

identity

Other

Institutional

identity

4.86

0.73

5.32

1.05

10.72 5.13 3.50 3.15 2.97 3.16 2.37 4.07 4.38

2.82 1.12 0.63 1.87 1.67 0.72 1.30 0.57 1.34

4.57 5.71 4.82 5.32 3.59 5.03 3.11 1.89 4.25

2.46 1.76 0.61 0.60 0.61 1.64 0.55 0.25 1.06

institutional identity

PNV-EA CIU CC PRC PR URAS

8.75 6.98 11.49 1.92 4.92 1.78

2.26 6.04 5.03 3.55 3.49 3.11

5.97

3.91

Table 6 lists salience values for each of the state-wide Spanish parties on the regionalist issue. It shows that Spanish state-wide parties devote on average roughly 5% of their party manifestos to the institutional aspects of the regionalist issue. The identity aspects receive even less attention and have a salience score of barely 1%. However, the table also makes

16

clear that state-wide parties do not systematically devote more attention to the regionalist issue in regional than in state-wide elections. The differences between PP and PSOE are small on average. While the PP appears to be a bit more reticent than the PSOE on the issue in statewide elections; in regional elections it is the PP which puts most emphasis on the issue, particularly with regard to its cultural dimension. Table 6 also highlights that in regions with the strongest presence of autonomist parties (type I), the regionalist issue is more likely to become salient. However, there are several outliers. For instance in Murcia, the PP makes the institutional dimension of regionalism more salient than expected, whereas it is not obvious why the regionalist issue should be much more salient in the Basque Country than in Catalonia, at least not on the basis of the strength of autonomist parties there. This said, on the whole, the salience that is attributed to the regionalist issue is lower than for the regionalist parties listed in the far-right column. Yet also in this regard, we observe significant outliers: the PP ‘outclassed’ the PNV in the Basque Country in terms of salience whereas the regionalist parties in Asturias and Cantabria have rather low values on the institutional dimension. Furthermore, it is also worth noting that most of the regionalist parties are not single issue parties in a sense that they do not focus on the regionalist issue alone. In fact, they devote on average 5.97% of their manifestos on institutional matters, which is only slightly above the average values reported for the statewide parties. But on closer look, there are considerable differences between the regionalist parties. The strongest regionalist parties (CC, PNV-EA, CIU) devote a much more substantial part of their manifesto to the issue (9.1% on average) than the smaller ones (2.9% on average; PRC, PR, URAS). In a sense, this confirms the characterization of some of these parties as nationalist and others as simply ‘regionalist’ i.e. parties which are able to make use of the opportunity structures that arise from the presence of regional elections. This difference is less outspoken with regard to the cultural dimension. The regionalist parties devote on average 3.91% to identity matters, which is considerably more than the state-wide parties. However, the Basque regionalist parties do not put more emphasis on identity matters than the statewide parties. Although we cannot infer the strategy of a party on the basis of salience alone, we find no evidence here of ‘less divergence’ in terms of salience on the regionalist issue for the PP than for the PSOE. This contradicts what could be assumed according to Hypotheses 4, given the more decentralized structure of its party organization. It is worth considering the salience profiles for both parties in somewhat greater detail. The slightly higher mean value of the PP on the institutional dimension than the comparative values for the PSOE is largely due to the Basque Country, where the party devoted no less than 10.7% of its manifesto to institutional issues (compared with only 4.57 for the PSOE) and 2.8% on identity matters (compared with 2.46 percent for the PSOE). The Basque PSOE also devoted more attention to the regionalist issue than in the other regions, but less spectacularly so than the PP. The regionalist issue also features prominently in the Catalan manifestos of both state-wide parties, even though the focus of the PP here is mainly on the institutional dimension. In both Murcia and Castile and Leon, the PSOE in particular applies an evasive strategy and downplays the issue (in both of its dimensions). The PP shows a similar but less consistent tendency. There is a relatively low emphasis on the issue, except in Murcia where the PP seems quite concerned with institutional matters.

17

In the four other regions (Asturias, La Rioja, Cantabria, Canaries) the most consistent pattern is that the PSOE is putting significantly more emphasis on the institutional aspects of regionalism than the PP. Compared with the other regions, the PSOE’s concern with the issue is particularly strong in Asturias and Cantabria, that is, the two regions where the PP is under threat from a regionalist party. Hence we could see this as a first potential indication of a strategy aiming at priming the issue in order to harm the state-wide competitor. In these four regions, the PP adopts a relatively low salience profile with regard to institutional matters. But in La Rioja and Cantabria, it puts relatively much emphasis on identity matters. Table 7 considers the salience of the regionalist issue in the UK. We first observe that the salience pattern in the UK shows a similar pattern. As in Spain, at least for the elections under consideration, the regionalist issue has not been very salient. Furthermore, like in Spain the institutional component of regionalism has been emphasized more consistently in state-wide and Scottish elections than its cultural component. In Wales however, the Conservatives ran against this trend by emphasizing regional culture more than the institutional dimension of regionalism. Even so, the institutional component of regionalism is not necessarily emphasized in Wales less than in Scotland or in state-wide elections. Overall, the state election programmes devote less attention to the issue than the regional manifestos, which is different from the Spanish case. Table 7. Salience of the regionalist issue: United Kingdom Salience (institutional/ cultural) Labour inst identity General 1997 2.47 0.26 General 2001 1.39 0.05 Scotland 1999 5.38 1.75 Scotland 2003 3.6 1.06 Wales 1999 5.33 0.67 Wales 2003 3.07 0 Mean all elections 3.54 0.63

Conservative Liberal Democrats inst identity inst identity 1.5 0.39 0.55 0 0.59 0 1.68 0.15 2.45 0.82 4.42 0.67 4.11 0.79 1.6 0.30 4.39 7.32 1.46 1.10 1.72 2.01 4.5 0.16 2.46 1.89 2.37 0.40

SNP Plaid Cymru inst identity inst identity 16.56 1.84 12.71 1.83 12.17 1.78 17.72 1.32 10.57 1.08 7.61 0.82 7.11 2.30 8.98 0.75 11.73 1.38 11.63 1.55

On the other hand, in terms of salience, the Liberal Democrats (as the organizationally most decentralized party and the state-wide party which most strongly favours a federal Britain) is certainly not the most salient on the regional issue. Also, if we leave aside institutional values for the Scottish elections in 1999 and Welsh elections in 2003, there is little variation in emphasis on the regionalist issue for the Liberals. It is worth considering some of these salience profiles in greater detail. Labour is most salient on the institutional item, while on average the Conservatives focus more on the identity theme (due to the high scores in Wales). The relatively high salience of the regional issue (in both dimensions) among the Welsh Conservatives coincides with a period in which the Welsh Assembly was only nascent and the Conservatives portrayed it as a quango (quasi nongovernmental organization). It could also offer a first indication for an ‘adversarial’ strategy, caused to harm the Labour Party (a similar strategy however was not pursued in Scotland, where Labour nonetheless also competes against a left-wing regionalist party). By 2003 however, the Welsh Conservatives ‘kept quiet’, and the Liberals were more adamant in supporting full legislative powers for the Welsh Assembly. Predictably, Table 7 also reveals that the autonomist parties devote considerably more attention to regionalism, especially on the institutional dimension.

18

Directional Certainty of the regionalist issue in Spain and the UK Table 8. Directional Certainty: Spain Type of Region

I II III

IV

PP Directional Certainty

PSOE

institutional

Identity

Other

institutional

State election 2000

-0.07

-0.35

0.08

Basque Country Catalonia Canary Islands Cantabria La Rioja Asturias Castile and Leon Murcia Mean regions

-0.02 0.28 0.07 0.25 0.38 0.30 0.25 0.31 0.23

-0.20 0.29 0.17 0.57 0.42 0.45 0.46 0.55 0.34

-0.07 0.73 0.62 0.40 0.42 0.39 0.44 0.51 0.43

identity

Institutional

identity

0.13 -0.33 PNV-EA 0.62 CIU 0.65 CC 0.57 PRC 0.38 PR 0.54 URAS 0.62 0.75 0.47

0.91 0.87 0.68 0.40 0.87 0.67

0.70 0.78 0.49 0.69 0.63 0.76

0.73

0.67

The directional certainty values allow us to make clearer predictions, especially with regard to the third Hypothesis. We assumed that state-wide parties are likely to be consistent in the direction on the regionalist issue, i.e. the stronger they take stance in state-wide elections in one direction, the less likely they will take stance in the opposite direction in regional elections. However, in the Spanish case, we observe that PP and PSOE hardly take a clear directional stance in state-wide elections on either dimension of the regionalist issue. State-wide parties in general elections have values close to zero, particularly with regard to the institutional issue (-0.07 for PP and 0.08 for PSOE). This means that it is hard to deduce on the basis of the regionalist sentences in the party manifesto alone which direction either party prefers, even though, as expected, the balance slightly tilts towards the regionalist side for the PSOE and the centralist side for the PP. The parties are somewhat less ambiguous with regard to identity issues. The PP clearly takes a more Spanish nationalist stand in its manifesto (-0.35) while PSOE slightly tends towards promoting the regional identities (0.13). In comparison to the manifestos for state-wide elections, the regional manifestoes of both PP and PSOE contain more cues in a regionalist sense, both with regard to institutions and identity politics. The overall mean of the state-wide parties is 0.33 on the institutional dimension and 0.41 on the identity one. In general, both PP and PSOE take a firmer regionalist stance with regard to cultural than institutional matters, that is, they are more restrained in supporting more regional autonomy than in advocating the identity of the region. Another consistent pattern which emerges from the regional manifesto data is that the PSOE takes a more distinctive stance in the regionalist sense than the PP, both with respect to identity and institutional matters. PSOE has an average score of 0.43 on institutional and 0.47 percent on identity matters versus respectively 0.23 and 0.34 for PP. However, the Basque Country forms a major exception to the general rule. This is most evident for the Basque PSOE: on both counts the directional balance of this party shifts to the centralist direction, especially so with regard to identity issues. With respect to the latter, the party takes an even more explicit centralist stand than the PP which is even slightly more ambiguous than in its state-wide manifesto. However, leaving aside the deviant Basque case, the pattern is remarkably consistent across the various regions. The tendency of the PSOE to take a more regionalist stand compared with state-wide elections is most outspoken in the Canary Islands, Catalonia and Murcia. In Murcia

19

also the PP takes a particularly explicit regionalist stance on both dimensions. On identity matters, however the PP is at its most radical in Cantabria and on institutional matters in La Rioja. Setting the Basque case apart, the PP is most ambiguous (and hence also closest to its position in the state election) in the Canary Islands. Finally, as for salience we have also looked at the directional certainty of the leading regionalist parties. Unsurprisingly, each of them features very high scores on the directional certainty variables (on average 0.73 on institutional matters and 0.67 on cultural matters), indicating that that they unequivocally opt for devolving more competences to the regions and promoting regional identity at the expense of national identity (table 8). The only exceptions are the institutional position of the PRC and the cultural position of the CC, which are more blurred compared to the other regionalist parties. The most ‘nationalist’ parties provide the clearest cues to the voters, particularly on the institutional dimension. Amongst the nonnationalist parties, it is only the PR in La Rioja which is equally outspoken with regard to regional autonomy.

Table 9: Directional Certainty: United Kingdom Labour inst

State election 1997 State election 2001 Scotland 1999 Scotland 2003 Wales 1999 Wales 2003 Mean

cult

Directional certainty Conservative Liberal Democrats inst

cult

inst

Cult

SNP inst

0.69 0.35 0.14 0.46 0.44 0.32

0.17 0.5 0.5 0.55 0.25 0

-0.37 -0.3 -0.13 -0.04 -0.5 0.33

0.30 0 0.5 0.1 -0.17 0.57

1 0.84 0.48 0.63 0.63 0.91

0 0.5 0.5 0.5 0.83 0.5

0.98 0.95 0.88 0.94

0.40

0.33

-0.17

0.22

0.75

0.47

0.94

cult

0.5 0.47 0.88 0.75

0.65

Plaid Cymru inst

cult

0.99 0.87

0.5 0.56

0.77 0.95

0.5 0.65

0.90

0.55

Compared with Spain, most of the British state-wide parties are clearer in their preferred direction with respect to the regionalist issue (the certainty values are higher) While the certainty of the autonomist parties is extremely high, the preferences of the state-wide parties are not hidden either in most of the cases. Except for Wales 2003, the Conservatives turned against institutional devolution in each of the other elections. In contrast, they evaluate regional identity and culture more positively. Only the result for Wales in 1999 is negative. This means that the Conservatives changed their strategy between 1999 and 2003 in Wales completely, going from a pre-centralisation viewpoint to a relatively clear stand in favour of at least the status quo. In other words, we find some evidence here for the H2, insofar as the Conservatives seek to harm the electoral interests of the Welsh Labour Party by emphasizing the regionalist issue (see higher) and taking a clear ‘centralist’/adversarial stance. The Liberal Democrats are giving clear cues what concerns institutional devolution, but are less clear when talking about identity. This is also true for Labour, although the certainty values are lower. The UK results also support the assumption that parties that are known to be most in favour or against extending devolution also express this unambiguously in their regional party manifestos (highest positive values for the Liberal Democrats, highest negative values for the Conservatives). Directional Intensity of the Regionalist Issue in Spain and the UK . We can now compute directional intensity values by multiplying salience with directional certainly. Although -100 and + 100 are the theoretically possible outliers, even regionalist 20

parties for instance - whose scores can be used as a benchmark - remain far below the theoretical upper limit of + 100 In Spain, the regionalist parties obtain an average score of 4.7 on the institutional dimension and 2.6 on the identity dimension. However, the differences between the autonomist parties are considerable, particularly on the institutional dimension. While CC (7.77), PNV-EA (7.96) and CiU (6.04) have a relatively strong regionalist profile, URAS (1.19), PR (4.27) and PRC (0.77) provide much weaker cues to the voter. The differences are smaller on the identity dimension, as the profiles of the most ‘nationalist’ among the regionalist parties are less intense than that on the institutional dimension. Considering the values for the state-wide parties, we notice that PP and PSOE adopt an extremely low profile on the regionalist issue in the state election. On both cultural and institutional matters, PP and PSOE provide weak cues respectively in a centralist and regionalist sense. Their stance is somewhat stronger with regard to institutions than culture. Table 10. Directional Intensity: Spain Type of Region

I II III

IV

PP Directional Intensity

PSOE

Institutional

identity

Other

Institutional

State election 2000

-0.34

-0.26

0.42

Basque Country Catalonia Canary Islands Cantabria La Rioja Asturias Castile and Leon Murcia Mean regions

-0.22 1.45 0.23 0.80 1.12 0.93 0.59 1.27 0.77

-0.56 0.33 0.11 1.07 0.70 0.32 0.59 0.31 0.36

-0.33 4.16 2.99 2.15 1.49 1.94 1.37 0.97 1.84

Identity

institutional

identity

0.14 -0.81 PNV-EA 1.10 CIU 0.40 CC 0.34 PRC 0.23 PR 0.88 URAS 0.34 0.19 0.33

7.96 6.04 7.77 0.77 4.27 1.19

1.58 4.69 2.46 2.45 2.20 2.37

4.67

2.62

In the regional elections both parties tend to shift towards a more regionalist profile, but this tendency is most outspoken on the institutional dimension. The PP switches from a moderately centralist stance (-0.34) to a moderately regionalist stance (0.77 on average). The PSOE intensifies its regionalist stance with regard to the institutional dimension and moves from 0.42 to an average score of 1.84. Thus, in the aggregate, the difference between the two state-wide parties remains considerable. As we have seen, this shift is largely due to a change in the substantive positions of the parties, in the sense that they become more explicitly proregionalist, rather than because they put more emphasis on the regionalist issue. With regard to identity issues, the differences between the state-wide and the (average) regional profiles are smaller. The PP switches from a weak centralist stand (-0.34) to a weak regionalist stand (0.36), while the PSOE sharpens its regionalist profile only slightly (from 0.14 to 0.33). The result is that in regional elections both state-wide parties have a highly similar profile on identity issues. At the same time, these averages mask a considerable variation between the regions. The differences between the regional party profiles are especially pronounced on the institutional dimension. Also, the latitude for strategic manoeuvring on this dimension is substantially higher for PSOE than for PP. We already know that this is mainly due to a larger interregional variation with regard to the substantive position of the PSOE on the regionalist issue.

21

The extent to which PSOE drifts off towards the regionalist end appears to be related to the strength of the regionalist competitor. In other words, the strategy of the PSOE becomes less evasive, the stronger the regionalist competitor (contrary to what was assumed in Hypothesis 1b). With the exception of the Basque Country, PSOE adopts the weakest profile in the regions without a regionalist party. Its profile is most intense in regions with strong regionalist parties (Catalonia: 4.16; the Canaries: 2.99) and somewhat weaker in regions with moderately strong regionalist parties (Cantabria: 2.15; Asturias: 1.94). At the same time it has to be kept in mind that the stronger stand of the PSOE in these regions is largely due to a qualitative change, as described above. Apparently the presence of a major regionalist party pushes the PSOE’s to take a more explicit regionalist stance, but not to put more emphasis on the issue. Furthermore, the shift towards regionalism does not appear to depend on the strategic position of the PSOE vis-à-vis the regionalist party. In Asturias and Cantabria, the competition for regionalist votes is largely between the PP and the more conservative regionalist parties, while in the Canaries and Catalonia the regionalist parties are more centrist and hence also threaten the PSOE. Yet in each of these cases the PSOE intensifies its stand. Therefore, in the case of the PSOE we find no evidence for varying strategies that are linked to varying positions of the regionalist competitors on the left-right axis (Hypothesis 2). Finally we should note that the strong profile of the PSOE is not always reflected on identity issues. In Asturias and Catalonia, the party takes a relatively clear stance in favour of promoting regional identity, but in the Canaries and Cantabria the party is much less committed in that respect. Contrary to the PSOE, there is not a single region in which the PP ventures far onto regionalist land. Its institutional stand varies much less between regions than in the case of PSOE (confirming Hypothesis 4) Variations in directional intensity on regional institutional issues are also less related to the strength of the regionalist parties. The party is most intense in Catalonia, but keeps very much aloof in the Canaries, in spite of the dominant autonomist CC. In Asturias and Cantabria, where the party faces a conservative regionalist competitor, its score is only slightly above the mean. Finally, the party adopts its second strongest profile and even surpasses the PSOE is Murcia of all places; a region in which an autonomist party is completely non-existent. The PP is even more prudent with regard to the issue of promoting regional identity. Even so, it sometimes leapfrogs a generally low-profile PSOE. This is most clearly the case in La Rioja and Cantabria. Again, there does not appear to be a relationship between the profile of the party and the competitive position of the regionalist party. This is best illustrated by the fact that, in the two regions where this competitive position is similar (Asturias and Cantabria), the PP adopts a fairly different profile. In the Basque Country, both PP and PSOE adopt a very distinct profile. In contrast to their strategy in all the other regions, they take a very cautious centralist stand, both with regard to institutions and identity. As mentioned above, this low profile is largely due to the ambiguity of the positions taken. Apparently, the salience of the ethno-national cleavage in this region compels both parties to allocate relatively much emphasis on the issue, while they do not want to commit themselves clearly to either direction. As a result the entrenchment between Basque and Spanish nationalists is not reflected in a clear-cut opposition between centralist and regionalist parties, particularly because the PP, contrary to what could be expected does not unequivocally play the Spanish card.

22

On the basis of the data we can now attempt to fit the investigated regions into the typology presented in Table 1. Scenario 10 would seem to apply to Murcia (type IV) and Castile and Leon (Type III), In these two regions a significant regionalist party is lacking and both state-wide parties opt for an evasive strategy. Apparently, in neither of both regions does PP or PSOE attempt to draw votes on a regionalist basis and claim the issue. Still, it is noteworthy that even in these regions, the parties shift substantially towards the regionalist pole. They take a relatively clear-cut stand in favour of promoting regional identity, but hardly emphasize this. Scenario 4 appears to have materialized in the cases of Asturias (type III), Cantabria (type II) and the Canary Islands (type II). Each of these regions has a significant regionalist party (type II regions in particular), and, with respect to the institutional dimension of the regionalist issue a more evasive (PP) and a more accommodative (PSOE) state-wide party. With regard to identity, both state-wide parties generally keep fairly aloof in these regions, expect for the PP in Cantabria (and to a lesser extent in La Rioja) and the PSOE in Asturias which put some stress on a fairly explicit regionalist position. The Basque Country (Type I) would seem to fit scenario 6, with a regionalist party and two low-profile state-wide parties. On closer look, however, the state-wide parties do not particularly de-emphasize the issue. The PP even puts an exceptionally strong stress on the issue. Yet, neither party makes an unequivocal choice for either the adversarial or the accommodative strategy. The scheme in Table 1, based on Meguid’s conceptual framework, does not take this possibility into account. Finally, Catalonia (Type I) appears to illustrate the first scenario. Confronted with a strong regionalist competitor, both state-wide parties go at some length to accommodate the claims for more autonomy. Even so, the PSOE goes much further than the PP, which remains relatively ambiguous. And both parties, but particularly the PP, are much less committed with regard to identity politics. In sum, the data strongly suggest that the regionalist issue generally functions as a valence issue. No party was found to opt for an adversarial strategy and take an intense centralist stand on the issue. Parties either adopt an intense regionalist or an evasive/ambiguous profile. We can now present data for Britain, and compare these with the Spanish directional intensity values. The British results are summarized in Table 11. Table 11. Directional Intensity: United Kingdom

General 1997 General 2001 Scotland 1999 Scotland 2003 Wales 1999 Wales 2003 Mean all elections

Directional intensity Labour Conservative Liberal Democrats SNP inst identity inst identit inst identity inst identity y 1,71 0,04 -0,55 0,12 0,55 0 16,3 0,92 0,49 0,03 -0,18 0 1,41 0,08 11,52 0,84 0,74 0,87 -0,33 0,41 2,14 0,34 9,35 0,95 1,64 0,58 -0,16 0,08 1 0,15 7,15 0,62 2,33 0,17 -2,2 -1,22 0,91 0,91 0,99 0 0,57 1,15 4,11 0,08 1,32 0,28 -0,48 0,09 1,69 0,26 11,08 0,83

23

Plaid Cymru inst identit y 12,59 0,92 15,44 0,73

5,47 8,49 10,50

1,15 0,49 0,82

In contrast with Spain, one party (Conservatives) has dared to adopt an adversarial centralist strategy, if only in Wales. Table 11 also reveals that for all state-wide parties, the intensity scores are more widely spread than for the Spanish parties. Of course, this comparison cannot be absolute since we consider more regional elections in Spain, but related to fewer electoral cycles (only one state-wide and one regional election) Taking a closer look at the specific issue profiles of the state-wide parties, we notice that, in the case of Labour, the lower directional intensity for institutional regionalism is more a consequence of the satisfaction with the institutional status quo than of a lack of emphasis. In every election Labour put more emphasis on institutional regionalism than on culturallinguistic issues. Positional values close to 0 on the institutional dimension do not come as a surprise: after all, Labour shaped devolution from the top (parliamentary acts) and from the bottom (by dominating the composition of all regional governments). Therefore, it would be odd if the party were to propose drastic changes to the devolution settlement, so early after devolution was established. Rather, with the competition from autonomist parties and a more proportional electoral system, the Labour party seems to have chosen to become a little more regionalist in terms of supporting Scottish or Welsh identity (the directional certainty on this dimension is consistently superior at 0.5). In line with our expectations (Hypothesis 4), the Liberal Democrats are the state-wide party with the highest directional intensity for the institutional dimension of regionalism. However, this is more a product of certainty than of salience. The 2003 Welsh manifesto is the manifesto with the highest directional intensity of any state-wide party. With regard to institutional change, the party remains committed to a federal United Kingdom and this is reflected in the party's scores. As indicated above, in the 2003 Welsh election the Liberals adopted their most intense position on institutional issues. Certainty values were higher than for any other state-wide party in the regional elections (for Wales and Scotland). As expected, the only party to have adopted a centralist position (negative score) on the institutional dimension is the Conservative party. It did so in each and every election but the last Welsh election. This position is consistent with the party's unionist creed and its initial opposition to devolution. At the same time, the intensity of the party's position is usually low. The most striking case of deviation from the central party is that of the Welsh party. In 1999, the party was more intensely centralist on both dimensions of regionalism than in any other elections. We can see this as an example of a adversarial centralist strategy aimed at harming the electoral fortunes of Labour and the Liberal-Democrats in their competition against Plaid Cymru (H2) However, in 2003, the Tories adopted a rather regionalist position. They apparently decided to market themselves as a more 'Wales-friendly' party to reflect the growing acceptance of (or adaptation to) devolution within the party. Hence the Welsh Conservatives retracted from their adversarial strategy by 2003. Taking such a clear negative stance against Welsh devolution was perceived as a risk. The relatively strong stance against devolution might have harmed Labour in 1999, but it clearly did not draw votes to the Conservatives either, as the party performed poorly in the election. In comparison, the Scottish Conservatives have been less reluctant to campaign against devolution throughout (negative score for directional certainty, but values for institutional regionalism are close to 0). However, save for the 1999 Welsh election, the Conservatives have shown a moderately favourable stance towards issues related to regional culture. A final observation is that all the state-wide parties have been relatively consistent in their regionalist profiles. The regionalism is only slightly more prominent in regional than in state-

24

wide elections, in particular when comparing the salience of regionalism between state-wide and autonomist parties. How can we fit the UK data in the typology developed in table 1? Scotland 1999 fits category 6 what concerns the institutional and identity dimension and the same can be said of the identity issue in 2003. While Labour and the Conservatives try to evade this issue as much as possible, the Liberal Democrats opted for the regionalist approach. On the other hand, on identity issues all state-wide parties adopt the evasive strategy when competing with the SNP in 1999 and 2003. In 2003 what concerns the institutional dimension all parties except for the conservatives who continue evading the issue, stress the regionalist issue (although the Liberal Democrats more moderately than in 1999). As such, this case fits category 4. In Wales, as mentioned above the Conservative party clearly chose an adversarial strategy in 1999, both on the institutional and the identity dimensions. At the same time Labour adopts a regionalist strategy on the institutional issue and an evasive strategy what concerns identity, while the Liberal Democrats opted for an evasive strategy on both dimensions. This places Wales 1999 in category 5 on both dimensions. In 2003 the Conservatives change strategy and start to evade the issue, while the Liberal Democrats draw the regionalist card on the institutional dimension and also Labour adopts an evasive strategy. On the identity dimension the Conservatives even go one step further in the direction of supporting the regionalist issue, while Labour and the Liberal Democrats choose to evade the topic. Both cases can be placed in category 6 for these reasons. 6. Concluding Observations First, both in Spain and the UK state-wide parties display moderate levels of divergence in terms of their salience profile, i.e. the extent to which they emphasise different issues in different regional elections. In the UK this divergence did not increase between 1999 and 2003. Whereas in the UK, state-wide parties tend to give more attention to the regionalist issue in regional elections as compared to general elections, there seems to be a reverse tendency in Spain as state-wide parties generally pay a little more attention to the issue in general than in regional elections (contradicting Hypothesis 1a) This difference can be attributed to institutional asymmetry. In the UK, the issue of whether Scottish, Welsh or Northern Irish autonomy should be extended concerns only a relatively small part of the electorate whereas in Spain regional autonomy affects the entire electorate. Second, although ‘regionalism’ features more prominently in Spanish general elections than in regional elections, it is also dealt with in a more ambiguous way. In regional elections the Spanish state-wide parties are more explicit in their policy choices, mostly in an autonomist direction. They may face the competition of autonomist parties and exclusively address the electorate of the region. Compared with general elections, ambiguity is smaller with regard to the institutional and identity components of regionalism. However, the difference with state-wide elections is most substantial with regard to identity politics. In sum, state-wide parties take a more intense regionalist stance in regional elections, not by increasing the salience of regionalism in their manifestos but by expressing themselves in a more straightforward way. A similar tendency was found in the UK, especially with regard to regional identity matters. Overall, the results for Spain and the UK reveal that state-wide parties are less restrained in advocating the identity of a region than in supporting more regional autonomy. Hence, state-wide parties counterbalance a moderately or ambiguously regionalist profile on the institutional dimension with the assertion of a strong identification

25

with the region; a logical outcome since the political consequences of supporting regional identity and culture for the state-wide party are small, whereas demands for more regional autonomy generate immediate political implications. Third, state-wide parties in both countries are rather consistent in their profile on the regionalist issue. A party which advocates a strong centre in general elections is very unlikely to advocate high levels of regional autonomy in regional elections (confirming Hypothesis 3) An exception to this pattern was found in the Conservative party in Wales, albeit that the party swapped sides between regional elections in 1999 and 2003 (and not between state-wide and regional elections). In the 1999 Welsh Assembly elections, the Conservatives strongly supported the Union, both with regard to institutional matters and identity politics. In 2003 however, the Welsh Conservatives adopted a fairly intense regionalist position, marketing themselves as a more ‘Wales-friendly’ party. This U-turn was most conspicuous with regard to promoting the Welsh cultural identity. But it also involved a more favourable stance with respect to regional autonomy, notwithstanding the more sceptic position of the Conservatives in both the 1997 and 2001 general election. The same ‘inconsistency’ albeit to a much lower degree has been displayed by the Spanish PP (and on this occasion, differentiating its strategy in state-wide from regional elections). In state-wide elections they adopt a low intensity profile in favour of a strong centre, but in the regional elections they generally take a somewhat more regionalist position, though always with a large dose of ambiguity and without giving the issue much salience. Perhaps, we can ascribe the inconsistency of the Welsh and Spanish Conservative parties to the dynamics of regional party competition which forces them into a more accommodative position with regard to the regionalist issue. Fourth, the analysis demonstrates that for all state-wide parties, state-wide party ideology functions as an anchor that prevents their regional branches from drifting off too far from the state-wide positions on regionalism (cf. H3). Thus, regional PP branches may tend to cross-over to the regionalist side in regional elections, but they maintain a very low intensity – in line with their more centralist creed - and almost never surpass the PSOE which is ideologically more inclined towards decentralisation. In the UK, the state-wide party that dares to go farthest in adopting a high-profile regionalist position in regional elections is the Liberal Democrats, but this corresponds with the more federalist ideology of the party. Fifth, it appears to be very difficult to predict the profile of the state-wide parties on the regionalist issue on the basis of their position on the left-right axis and the expected competitive dynamics in the system (disconfirming H2). The only systematic relationship found in the data was that, in Spain, the PSOE takes a less ambiguous stand in favour or regional autonomy, the stronger the autonomist party in the regional system. This pattern is actually more in line with the spatial model of party competition: the stronger the regionalist party, the more the centre of gravity in the system will move towards the regionalist pole of the dimension. We may assume that this will also have an effect on public opinion in the sense that the distribution will shift to the regionalist pole and the modal voter will become more autonomist. From a spatial point of view, it is logical that the state-wide party will take suit and also shift to the autonomist pole. At the same time, the fact that the state-wide parties never allocate a lot of emphasis on the issue and thus avoid an intense position on the issue attests to the directional point of view that, in a multi-party system, the parties will not try to beat the regionalist parties on their own terrain, but will rather attempt prime their own core issues and make these more central to the voter’s choice. On the other hand, whether an autonomist party competes on the left, right or the centre does not appear to make a difference. Nonetheless, we predicted that Conservative state-wide parties (being ideologically more inclined towards defending the centre) would have opted for an intense centralist stand in regions where the socialist state-wide party competes against a

26

left or centre-left autonomist party. By increasing the salience of the regionalist issue, such an adversarial strategy could harm the electoral support for the targeted socialist party (although these votes would benefit the autonomist party, not the Conservative competitor). However, the only conservative party which appears to have attempted such a strategy were the British Conservatives in the Welsh 1999 election. Even though this adversarial and unionist strategy may have harmed Labour in 1999, four years later the Welsh Conservatives retraced from this strategy and, as mentioned, adopted a more Wales-friendly profile, both with regard to institutional devolution and identity politics. Conversely, in some Spanish regions, the autonomist party has a more right-wing profile and competes mainly with the conservative PP. Following the same ‘adversarial logic’ of above, in these instances, we expected the PSOE to raise the salience of the issue – most plausibly in a regionalist direction – thereby seeking to harm the electoral support for the PP. However, to our surprise, the PSOE did not attempt such a strategy in these cases. By and large, most state-wide parties shift slightly towards the regionalist pole in regional elections, but purposefully retain a ‘low intensity’ profile. Instead, they will seek to prime their own core issues and make these more central to the voter’s choice; a finding that is consistent with directional theory. Therefore, notwithstanding some intra and inter-party differences, the evasive strategy remains the most dominant strategy of the state-wide parties, as predicted under H2. However, it is important to consider the combined effect of salience and position (directional certainty). Particularly in Spain, the main difference between the regional and general election profiles is not that the state wide parties either differentiate between the emphasis they put on the issue, or advocate different policies, but rather that they are more equivocal as to the preferred direction, as a result of which the voter can be more certain about the party’s position. The theoretical consequence is that a simple typology which differentiates between an evasive, adversarial and accommodative strategy (as proposed by Meguid) may be insufficient to capture these different strategies. Arguably, one could limit the label ‘evasive’ to a party which does not give the regionalist issue much salience and use the label ‘ambiguous’ to capture the strategy of a state-wide party which makes the issue salient, yet remains close to the SQ or issues inconsistent preferences. Finally, in explaining the limited divergence between state-wide and regional party branches in terms of their issue profiles, we assumed that state-wide parties that are more centralized are less likely to tolerate territorially divergent campaign strategies than state-wide parties with a decentralized party organization (H4). We found some evidence to support this assumption, but only in the Spanish case. The PSOE is ostensibly more decentralised than the PP and, perhaps as a result, its profiles on the regionalist issue are the most territorially divergent. However, a similarly consistent pattern was not found for the UK Therefore, a more powerful predictor of intra-party variation can be found in the incumbency status. The larger strategic leeway which the PSOE branches appear to have enjoyed may well have been due to that party’s role in central opposition at the time of the 2001/2003 regional elections. This is consistent with similar findings in the UK where Labour - as party in central office – appeared to have tolerated less divergence in issue profiles than the Conservatives who were in opposition at both levels, while the Liberal Democrats (who were in office at the regional level alone) occupied an intermediate position. Bibliography Ashworth, R. (1999). Political Priorities: A Content Analysis of Welsh Local Election Pledges. Paper presented at the Annual Conference of the British Political Studies Association. http://www.psa.ac.uk/cps/1999/ashworth.pdf

27

Budge, I. and D.J. Farlie (1983), Explaining and predicting Elections: Issue-Effects and Party Strategies in Twenty-three Democracies. London, Allen and Unwin. Budge, I., D.Robertson and D.Hearl (eds.) (1987), Ideology, Strategy and Party Choice. Spatial Analyses of Post-war Election Programmes in 19 Democracies. Cambridge, Cambridge University Press. Budge, I., H.D. Klingemann, A.Volkens, J.Bara, and E.Tanenbaum (2001). Mapping Policy Preferences: Estimates for Parties, Electors, and Governments 1945-1998. Oxford: Oxford University Press. Buznego, O.R. (1998), Elecciones autonómicas, sistema de partidos y gobiernos en Asturias, in : M. Alcántara and A. Martínez (eds.), Las elecciones autonómicas en España 1980-1997. Madrid, CIS, 81-97. Díez-Medrano, J. (1995) Divided Nations. Class, Politics and Nationalism in the Basque Country and Catalonia. Cornell University Press. Downs, A. (1957), An Economic Theory of Democracy. New York, Harper and Row. Enelow, J.M. and M.J.Hinich (1984), The Spatial Theory of Voting : an Introduction. Cambridge, Cambridge University Press. Fabre, E. and Maddens, B. (2007), 'Party organisation in multilevel systems - different party responses to the regionalisation of politics', paper presented at the ECPR General Conference, Pisa, 6-8 Sept. Fabre, E., Maddens, B., Swenden, W. and Libbrecht L. (2006), ‘Organisation and Strategy of state-wide parties at the subnational level in the UK’, paper presented at the ECPR graduate conference, Essex, 7-9 Sept. Fernández Ferrero, M.A. (1997), Procesos electorales : elecciones autonómicas en La Rioja 1979-1995. Logroño, Universidad de La Rioja. Hernández-Bravo, J. (1998), La construcción electoral de Canarias en la autonomía : una questión no resuelta, in : M.Alcántara and A.Martínez (eds.), Las elecciones autonómicas en España 1980-1997. Madrid, CIS, 119-149. Llamazares, I. and F. Reinares (1998), Elecciones autonómicas y sistema de partidos en La Rioja (1983-1995), in: M.Alcántara and A.Martínez (eds.), Las elecciones autonómicas en España 1980-1997. Madrid, CIS, 119-149. Llera, F.J. (1998), Pluralismo y gobernabilidad en Euskadi (1980-1994), in : M.Alcántara and A.Martínez (eds.), Las elecciones autonómicas en España 1980-1997. Madrid, CIS, 413-443. Macdonald, S.E., O.Listhaug and G.Rabinowitz (1991), Issues and Party Support in Multiparty Systems. American Political Science Review, 85, 1107-1131. Macdonald, S.E. and G.Rabinowitz (1993), Direction and uncertainty in a model of issue voting. Journal of Theoretical Politics, 5, 1, 61-87.

28

Maddens, B. (1994), Kiesgedrag en partijstrategie. De samenhang tussen de beleidsmatige profilering van de partijen en het kiesgedrag van de Vlamingen op 24 november 1991. Leuven, Afdeling Politologie. Martín, E. (2003), Asturias, in : Oriol. B. (ed.), La competencia política en la España de las autonomías. El eje izquierda-derecha en las Comunidades Autónomas. Barcelona, ICPS, 7993. Meguid, B.M. (2004), The Critical Role of Non-Proximal Parties in Electoral Competition: Evidence from France. Paper prepared for delivery at the 2004 Conference of Europeanists, Chicago. Meguid, B.M. (2005), Competition Between Unequals : the Role of Mainstream Party Strategy in Niche Party Success, in : American Political Science Review, 99, 3, 347-359. Molina, M. (2003), Cantabria, in : La competencia política en la España de las autonomías. El eje izquierda-derecha en las Comunidades Autónomas. Barcelona, ICPS, 129-142. Mora, H. (2003), La Rioja, in : La competencia política en la España de las autonomías. El eje izquierda-derecha en las Comunidades Autónomas. Barcelona, ICPS, 315-332. Ocaña, Francisco.A. and Oñate, Pablo. (2000). ‘Las elecciones autonómicas de 1999 y las Españas electorales’, Revista Espaõla de Investigaciones Sociológicas, 90:183-228. Pallarés, F. (1994), Las elecciones autonómicas en España : 1980-1992, in : P.del Castillo (ed.), Comportamiento político y electoral. Madrid, CIS, 151-220. Pallarés, F. and M. Keating (2003), Multi-Level Electoral Competition. Regional Elections and Party Systems in Spain, in : European Urban and Regional Studies, 10, 3, 239-255. Pallarés, F., J.R. Montero, and F.J. Llera (1997), Non State-wide Parties in Spain. An Attitudinal Study of Nationalism and Regionalism, in : Publius : the Journal of Federalism, 27,4, 135-169. Petrocik, J. (1990), The Theory of Issue Ownership. Issues, Agendas, and Electoral Coalitions in the 1988 Election. Unpublished paper. University of California, Los Angeles. Petrocik, J. (1996), Issue Ownership in Presidential Elections, with a 1980 Case Study, in : American Journal of Political Science, 40, 825-840. Rabinowitz, G. and S.E.Macdonald (1989), A directional theory of issue voting. American Political Science Review, 83, 1, 93-121. Ramos Rollón, M.L. (1998), Análisis de las elecciones autonómicas en Cantabria : el difícil proceso de creación de una autonomía, in: M.Alcántara and A.Martínez (eds.), Las elecciones autonómicas en España 1980-1997. Madrid, CIS, 151-177.

29

Seawright, D. (2004), ‘The Scottish Conservative and Unionist Party: ‘the lesser spotted Tory?’’, POLIS Working Paper No.13, School of Politics and International Studies, University of Leeds. Soler, J. (1998), Las elecciones autonómicas en Cataluña (1980-1995), in: M.Alcántara and A.Martínez (eds.), Las elecciones autonómicas en España 1980-1997. Madrid, CIS, 225-256. Soriano, E. (2003), Canarias, in : La competencia política en la España de las autonomías. El eje izquierda-derecha en las Comunidades Autónomas. Barcelona, ICPS, 111-128. Volkens, A. (1992), Content Analysis of Party Programmes in Comparative Perspective: Handbook of Coding Instructions. Berlin, WZB. Webb, P. (2000), The Modern British Party System (London: Sage).

30