Secessionism as a party competition game: The radicalization of the self-determination demands of the ethno-regionalist parties in Western Europe

Transcription

1 Secessionism as a party competition game: The radicalization of the self-determination demands of the ethno-regionalist parties in Western Europe Author: Josipa Rizankoska PhD Candidate in Comparative and European Politics at CIRCAP University of Siena, Italy Address: Via Mattolini 10, Siena, Italia josipa.rizankoska@gmail.com Abstract The debunking of the architectonic illusion of the realism of the group has served as a scholarly motivation to no longer consider national minorities as internally homogeneous and externally bounded groups, but to move towards wider perspectives in the way they observe the territorial politics. In spite of the fact that national minorities parties, otherwise known as (ERPs), have been considered as very inflexible players, as well as mainstream parties strategies dependent, ERPs act in a multi-dimensional competition game and deserve to be treated as a flexible electoral players with its own electoral strategies and tools for vote maximizing. New theories address not only the outbidding competition within an ethnically heterogeneous society but also the competition beyond the segmented market. Therefore, the radicalization of the ERPs demands for self-determination can now be perceived from the party competition game perspective. This empirical comparative model is an attempt to make up for the lack of empirical Large-N proofs of the already existing theories on the radicalization of the ERPs demands for secession at national level. It will look at the radicalization phenomenon from perspective of intra and inter group party competition in the countries of Western Europe. The findings suggest that when a certain ERP takes part of a regional government on the day of national elections the odds that this party will act as secessionist are lower. The mainstream parties positions turn to influence very little if any, and not in a constant direction for all the issues of interest for the ERPs. If on the day of national elections there is another ERP, from the same region, holding seats in the National Parliament, then the odds that another ERP will be secessionist are higher. When there is an electoral challenger from the same region the probabilities for radicalization get lower. This finding implies that the outbidding game towards radicalization is not the rule, and it opens space for the novel (nested competition game) theory to be tested in depth. Keywords: territorial politics, federalization, secession, ethno-regionalist parties (ERP), electoral competition. Introduction Nowadays, Europe abounds with regions where territorially concentrated groups assert aspirations for their right for national self-determination. (Miller, 1995) Some nationalist movements, like the one of Catalonia, are negotiating the right for independence referendum within Spain, i and strives towards sooner conclusion of its secessionist demands throughout democratic tools. For the Scottish independence movement in the UK, the final countdown has already started towards the opening of the referendum polls. ii Very often, the ethno-regionalist phenomenon has been investigated under the presumption that it drives potential for secession by violent means. Sorens (2012), nevertheless, has examined the methods and the causes of secessionist mobilization worldwide, as well as in the advanced democracies, to note that violent secession is a Page1

2 rare phenomenon in advanced democracies, and the struggle for independence normally goes with democratic-electoral tools. He does warn, though, that if gazing only in advanced democracies, one might develop a rosy view on secessionism. In democracies, Ethno-regionalist parties (ERPs) have arisen as main political players claiming to hold the ownership of the center-periphery cleavage (Lipset and Rokkan, 1967). They speak for certain group that collectively suffers, or benefits from, systematic discriminatory treatment vis-a-vis other groups in a society. According the restrictive definition used by Muller-Rommel (1994), ERP refer to the efforts of geographically concentrated peripheral minorities which challenge the working order, and sometimes even the democratic order, of the nation state by demanding recognition of their cultural identity. Consequently, one can see ERPs as anti-sentiment or even anti-system parties. In the same sense, Kitschelt (2011) notes that ERPs do not strive towards integration into broader national identities, but they limit their appeal to certain constituencies, and aim at securing better selfdetermination of the region. In spite of the debatable nature of the party family, there are different normative views on the future of this party family. Namely, Rasmussen (1991) stands on the position that ERP should stick to one theme the autonomy issue in order to assert maximal representation, whereas Montabes et al. (2004) are in favor of moderation of the autonomist issue position in order to obtain better electoral support (especially within proportional electoral systems). This party family has been often neglected in the literature mainly due to the small size and low turnout that the region performs on national level, due to the low political impact on national policy-making, and mostly due to lack of information (Muller-Rommel, 1994). In the past decades, nevertheless, ERPs have increased their size and additionally have managed to increase their blackmail potential and coalition bargaining power; thus, their policy-making influence has been increasing as well. Going back to the idea that we deal with a special party family whose ideologies by default challenge the existing political system (Elias and Tronconi, 2011) the concern arising from their rising importance in national politics does not come from their threatening capacity (their electoral strength), but from their threatening intention (their secessionist ideological position) (Massetti and Schakel, 2013). Consequently, the puzzle in this article originates from the threatening intention intensity and consistency, and the influence it has in the national electoral game. Understanding what accounts for the pro-secessionist self-determination demands of an ERP in the advanced democracies of Western Europe will be the main goal of this research. The debunking of the architectonic illusion of the realism of the group (Brubaker, 1998) has served as a scholarly motivation to no longer look at national minorities as internally homogeneous and externally bounded groups, where power-sharing model would fit the best (Lijphart, 1977, 1984; Norris, 2008), but to open horizons to new theories on the ERP role in the ethnic politics. Apart from the outbidding competition within ethnically heterogeneous societies (Rabushka and Shepsle, 1972; Horowitz, 2000), novel theory on the competition beyond the segmented market by Zuber (2012) has entered in the game of territorial politics. These theories no longer neglect the intra-group party competition; moreover, the radicalization of the ERPs ideology became more and more perceived as an intragroup (Bochsler and Szöcsik, 2013), as well as an intergroup electoral party competition game. (Meguid 2005, 2008; Elias and Tronconi, 2011) Uneven in scope and method, literature on ERPs has been usually based on descriptive, case study analyses. De Winter and Tursan (1998), and De Winter, Gomez- Raino and Lynch (2006) later offered a general analysis of many salient ERPs in eight countries of Western Europe. So far, the ERPs origins, motivations, typologies, electoral entry, success, persistence and failure, organization and leadership, coalition behavior and their positions regarding European Union have been addressed by the literature. (Coakley, 1992; Hooghe, 1992; Ishiayama, 2011; Brancati, 2007; Candra 2004, 2011 etc.) A few Large-N comparative analyses focusing on the ERPs entry, size and electoral success have been issued recently by Tronconi (2009), or Bernauer and Bochsler (2011) who managed to combine analyses of the contextual variables and variables related to the parties themselves. In spite of some attempts to position the parties on different ideological scales and to classify them according to their level of radicalization of their ideologies towards self-determination (Muller-Rommel, 1994; Dandoy and Sandri, 2007; Dandoy 2010; Protsyk and Garaz 2011; Gadjanova 2012, 2013), few Page2

3 comparative empirical analysis on the importance of the ideology of the party family have arisen. Massetti and Shakel (2013 a, b), have created a unique cross section-cross cross-time dataset classifying the ERPs according to their ideological radicalization towards territorial self-government. They run regression models on the ERPs ideological radicalization towards secession and try to explain the final outcome through decentralization level and decentralization reforms within a nation state. So far, this has been the best contribution in terms of empirical over time and space comparative data on the regionalist parties ideological extremism towards secession. iii The comparative model offered here is an attempt to make up for the lack of empirical Large N proofs for already existing theories on the radicalization of the ERPs demands for secession, as a result of party competition strategies at national level. The analysis is focused on the Western European cases of ERPs electoral campaigning in democratic elections after World War II. In the following sections of the article I will firstly introduce the debatable nature of the Ethno-regionalist party family itself, and the criteria taken for the case selection in this paper. Later I will define the hypotheses based on the theories from the existing literature on the competition factors for different ideological stances of the ERPs. The ERPs participation in national or regional governments; the competition tactics of the mainstream parties towards the niche issues of ERP s ownership, and the intra-group parties competition will be tested. I will then explain the model and the variables operationalization, to finally present the results and the conclusions of the findings. In appendixes, the robustness check analysis will be enclosed, as well as several tables on the dataset properties. The findings support my assumption that the ERPs are flexible strategic electoral players whose scope of issues is wider than the territorial self-determination card on bases of which the party family has been defined as such. The odds of finding radical parties lower when an ERP party had been participating in a government before national elections. The chances for radicalization rise when the two mainstream competitor parties apply adversarial tactics on the issue of multiculturalism, and when a party has an already assessed ERP competitor that took seats in the National Parliament before the day of elections. The higher parliamentary fractionalization and higher decentralization level are also significantly more fruitful soil for moderation than the higher economic development of a country which on the contrary boosts radicalization. The Ethno-Regionalist Party dataset: case selection criteria The initial methodological challenge when dealing with analysis on ERPs is the ongoing scholarly debatable nature of the party family. Many authors, who are trying to define the characteristics of the parties that constitute this party family, find the ideology as one of the crucial criteria for categorization and definition. To start with, the range of demands of ERPs, following Jenne s (2007) conceptualization, includes two frameworks of demandinside and outside of state. The inside of state framework goes on a threefold scale of radicalization of the ERPs demands for self-determination: affirmative action as least radical, then cultural autonomy, and finally territorial autonomy as the most radical. Secession or irredentism is the highest level of radicalization which belongs to the outside of state framework. Muller Rommel (1994) has introduced a four scale typology for the ERPs positions: the least radical- protectionist, the autonomist, the left-libertarian federalist and finally the highest level of radical - separatist. Dandoy s (2010) frame for analysis and comparison of ERP types has been based on ideology and, more particularly, on the essential dimension of the strength of the party demands regarding the preferred state structure and the future of their region. Three main types and several sub-types of ERPs have been identified: protectionist (soft demands), decentralist (mild demands with challenges to the internal order) and secessionist (strong or radical demands with challenges to the international order). Gadjanova (2012) has categorized the concepts of identity politics in electoral campaigns under the following groups: the community (own group), the group s past, group rights, means to safeguard group rights, discrimination/unequal treatment and other groups. Given that the territorial dimension arises in all of these, and in more other attempts to define and classify the party family, some authors had considered them as niche parties holding the ownership of the center-periphery cleavage (Meguid, 2005). Having in mind that ideology radicalization towards secession is the dependent variable of this analysis, I rely on two sets of criteria in defining the ethno-regionalist party family and thus, in the process of data gathering. The first one has been offered by Tronconi (2009) who Page3

4 points out two crucial features that a party must possess to be categorized within the family of ethno-regionalist parties. Namely, it should represent a certain ethnic (national) group, and the group should be territorially concentrated within the nation-state. The second set of criteria is based on Massetti and Shakel s (2013, b., pp.4) conditions: the party should be a self-proclaimed political organization which contests elections; it should present and/or field candidates only in a particular sub-territory (region) of the state (which corresponds with Tronconi s second criteria), and the territorial limitation of their political/electoral activity is a consequence of their explicit objective of defending the identities and interests of their region. Their fourth criteria has been driven by De Winter s (1998, pp.204) statement that regionalist parties core mission is to achieve/protect/enhance some kind of territorial selfgovernment of their homeland; therefore, this criteria excludes regional parties that are regional branches of mainstream parties and those that are only ethnic parties. This is because these parties, according to the two authors, are not interested in the territorial aspect of selfgovernment, but in community rights only. The data used for the dependent variables measuring radicalization of the ERPs has been contributed by Masseti and Shakel (2013a), on which more will be written later, and thus the case selection criteria relies on most of their criteria, with some extra details on included cases in the scope of this analysis. Firstly, the ethnic parties that represent a territorially concentrated group within a nation-state, and have a historic territory they refer to as a homeland, but do not explicitly want to self-administer the region(s) they live in, are also taken in the analysis classified as protectionist (e.g. The Sweedish People s Party SFP in Finland); secondly, the parties that have regionalist identity but their ethnic or nationalistic feelings are not the main driver of the regional politics have been excluded (e.g. Extremadura Unida and La Rioja in Spain); Moreover, the two Unionist parties in Northern Ireland are excluded due to their pro Unionist claims; however, Sinn Fein and SDLP are taken into consideration, despite their uncertain pro-autonomist or alternative border-drawing claims; finally, parties that are ethno-regionalist in all of the above mentioned criteria but, at one point in time have demonstrated radical-right parties features, (e.g. Vlaams Belang in Belgium, Lega Nord in Italy, etc.), have not been excluded because the internal ideological differences of the ERPs goes along with some of my main arguments, and it adds value to itself. Because of the ERPs usually small size or lack of representation in national legislatures (due to some electoral systems disproportionality), the selection criteria does not lay on the ERPs relevance measured as electoral success. The only exclusive criteria for a party complying with all the above presented criteria is the principle of campaigning at least for two national elections, with which criteria a party demonstrates certain continuity of its existence. In fact, only two French ERPs, Accolta Naziunale Corsa (ANC) and Alsace d'abord (ADA), cannot demonstrate two consecutive independent campaigns at national elections in this data set, but they can demonstrate a long existence at regional level and two national elections with some electoral support. Theory and Hypotheses Many studies on ERPs are treating this party family as having a special role of territorial politics (Keating, 2011), and focus mainly on country dependent variables, such as culture or economy, to explain the role ERPs had in the nation-state politics. Sorens (2012) argues that secessionist vote share in democracies can be explained by several factors: provinces with a history of independence, provinces with their own languages, provinces with multiparty political systems, and geographically separated from the rest of the country, etc. This article instead, is presuming normality of the territorial politics and shall contribute to the group of scholars that treat ERPs as vote-maximizers and officeseekers (Elias and Tronconi 2011, Zuber 2012, Bochsler and Szocik 2013), rather than zero-sum political actors (Keating, 2011). The hypotheses tested in the models here view the ideological positions of the ERPs in national elections as electoral-game based processes, rather than phenomenon dependent on nation-state factors. I depart this research exactly from the premise that ERPs are vote-maximizing electoral competitors of the mainstream, as well as the other ERPs, and are officeseeking rational players in the processes of government coalition bargaining. For this purpose they, very often, if not always, have to adjust their pro-secessionist rhetoric vis a vis their objectives within the complete plethora of electoral players and possible positions. Bocshler and Sozik (2013) claim that the inclusion of minority parties into national governments produces an inherent conflict between pragmatic office seekers and radical partisans. In minority parties that have governmental responsibilities, the pragmatist view overwhelms, whilst Page4

5 in those parties in opposition, radical voices dominate. So, they show that government inclusion, as the most important element of power sharing arguments, can destroy the political unity of the ethnic minorities and contribute to their radicalization. This stands in contrast to the classical ethnic-outbidding hypothesis, according to which within group competition is the main driver for radicalization (Rabushka and Shepsle 1972; Horowitz, 2000). Tronconi and Elias (2011) assess that the risks autonomist parties have to take when included in government is just as present in many coalition building practices that do not necessary involve ERPs. ERPs, whatsoever, lower the secessionist rhetoric when in government. This might occur due several reasons such as: their ambitions to appeal to larger electorate (as office seekers) by mainstreamizing the electoral platforms and abandoning radical stances; the coalition bargaining settlements can push the ERP to adopt a moderate ideology in order to better appeal to the mainstream partner electorate; and the public policy benefits that a party can have in government can motivate the ERP to offer a wider electoral issue platform, or at least, going back to the first one, to moderate the positions on the territorial issue in order to gain more votes, etc. (Sorens, 2012) I formulate the first hypothesis on the basis of the prevailing stances in the literature: H1: The ERP s inclusion in national or regional government decreases the probabilities for adopting radical electoral positions on the self-determination issue. Budge s salience theory (1992, 2001) on the other side, argues that parties play on some key issues and priorities in order to demonstrate that they are the credible and trustful owner of them. Meguid s theory (2005, 2008) assumes that when a mainstream party takes a dismissive strategy towards the territorial issue it signals lack of merit for the issue itself, but this salience reduction does not affect the ownership of the issue, i.e., the territorial issue stays in the hands of the ERP. Among several strategic scenarios within her Modified Spatial Theory, there are two in particular indicating how the two biggest mainstream parties can affect the ERP s electoral success. Firstly, if the two mainstream parties go for the accommodative strategy they do challenge the ownership of the ERPs, and as a result of given advantages that MPs normally have over the ERPs (such as more access to voters that allows them to publicize their issue positions and establish namebrand recognition), the ERP loses votes. In the opposite scenario, when both main mainstream parties decide to oppose the territorial issue, and adopt an adversarial tactic, the salience is high, but the ERPs still holds the ownership, and in this case does not lose votes. Following this logic, in the first case, where the ERP has a serious threat from the mainstream parties over the ownership of the salient issue I expect that it will affect the ERP in radicalizing in order to defend itself as the rightful issue owner. In the second case, where the adversarial tactic has been applied, the ERP has no need to radicalize itself as it is already the owner of the issue and in order to gain more seats it will have to moderate. (Montabes, 2004) The rational choice approach by Levi and Hecher (1985) argues that ERPs obtain protest votes due to their nature of offering public goods that promisis political change instead of generating private benefits to their voters. This is because ERPs, which usually have small size, have little power to offer career benefits to more than the party s elite members. They predict that ERPs are here to stay as long as the political issues of its followers remain on the political agenda and are not adopted by a major party. This theory implies that ERPs are not flexible to wider their scope of issues due to their necessity to maximize votes through the ownership of the ethno-regionalist issues. The rational choice theory in this case, by presuming ERPs inflexibility in creating their electoral profile does not meet the Kirchheimer s models (1966) that assert the progressive convergence of parties to more centrist positions within their party systems as well as retrenchment of traditional denominational cleavages salience. Elias (2013) also speaks of the other face of ERPs as vote-maximizing actors that have the same strategic tools as mainstream political parties and bases her argumentation on the premise that the ERPs are departing from the niche identity towards mainstream. Gadjanova s (2013) Ethno-regionalist Index (which will be more thoroughly introducd in the variable operationalization section) contains two main issues of a crucial importance for the ERPs, the territorial issue and the issue of multiculturalism. A positive trend of these issues on the Index scale should indicate higher ethnoregionalist identification. If these two issues considered to be signal of the intensity with which certain ERPs addresses the regionalist issue are taken as such, testing the competition that comes from the mainstream parties at national level shall be tested on these issues as well. That is to say, to test whether ERPs are inflexible and handicapped electoral players (as the Meguid s theory Page5

6 introduced before was suggesting) or they might adopt electoral tactics depending on the specificity of the issue and the ERP s own electoral strategies (introduced by Elias and Tronconi, 2011), I look at the mainstream parties stances on the two general ethno-regional issues at national elections and follow the reaction of the ERPs in terms of radicalization. I construct two hypotheses that shall test the connection between the radicalization of the ERPs ideology according to the two largest mainstream parties positions towards the main ethno-regionalist issues (the territorial and the multiculturalism one). H2: The accommodative tactic of the two largest mainstream parties on the regionalist issues will radicalize the ERP s appeal towards secessionism. H3: The adversarial tactic of the two largest mainstream parties on the regionalist issues will moderate the ERP s appeal towards secessionism. The other set of hypotheses addressing the electoral strategies refers to the intra-group party competition in relation to ideology change of the ERPs. Two general theories have been developed regarding the intra-ethnic group party competition strategies: the outbidding hypothesis and the nested competition game. The first one claims that competition between parties addressing the same ethnic group leads to a race to the extreme: each competitor will need to position itself as the most credible representative of minority interests, involving increasingly radical arguments. Recent studies, on the other hand, have shown that ethnic outbidding is not an inevitable result of competition within the same group (Candra 2005; Mitchell, et al. 2009; Zuber, 2012). In this direction, Zuber presents the theory of nested competition which is a party competition in an imperfectly segmented market where some, but not all, parties make offers across ethnic divides and where competition in intra-ethnic arenas is nested within an intra-ethnic arena of party competition. Thus, ERPs can choose exclusive or non-exclusive appeal to an ethnic identity category. Additionally, ERPs can choose between different policy positions on an ethnic dimension of competition. Nested competition therefore emphasizes the previously neglected fact that ethnic elites choose their strategies not only in light of societal preferences (outbidding) and institutional parameters (power-sharing), but also in light of the behavior of their competitors. In particular, if large mainstream parties which accommodate the demands of ethnic minorities are electorally successful, ethnic minority parties gain a signal that votes can be gained on the moderate end of the spectrum of minority demands. If ethnic parties begin to see mainstream parties as their primary competitors for votes, ethnic outbidding is unlikely to appear as an attractive strategy to regain votes willing to vote for the mainstream party. In line with these theories I offer three hypotheses: H4: The ERPs that are holding seats in the National Parliament in time of national elections, due to the logic of race to the extreme, shall boost the radicalization of the other ERP s appealing to the same ethno-regional group. H5: The representation of ERPs in National Parliament in general (no matter which region or ethno-regional group they represent) will moderate the ideology of the ERP, as an example of a successful votemaximizing/office seeking strategy. H6: The regionalist competition from another ERP at national elections is expected to drive radicalization, if the outbidding model holds, or to push towards moderation if the nested competition could be the new game in town. The control electoral competition variables will test the size and the electoral system representation provisions. It is expected that the size of the ERP improves the odds of profiting from parliamentary or governmental inclusion provisions, thus it shall moderate their electoral stance in order to maximize their votes. The more fragmented Parliament which in large inclines inclusive electoral systems should lower the radical stances of the ERPs due to the same reasons - better odds for representation. Hence, the hypotheses: H7: The larger the size of the ERP the less probability to have a secessionist party. H8: The more fragmented the national legislative body is, the less probabilities for secessionist parties there are. Decentralization as an explanatory variable for radicalization has been tested by Massetti and Shakel (2014b), to conclude that there is a strong positive direction between decentralization reform and radicalization. Nevertheless, the decentralization level is stable over regions (countries) over time, and can have many different effects compared to the decentralization reform. Negative direction could also be expected due to the possibility for better representation of the ERPs in the territorial multilevel politics in countries with higher level of decentralization (as implied by Montabes et.al. 2004). It could be expected that in countries with a higher degree of decentralization where parties Page6

7 participate in multilevel electoral politics ERPs are likely to turn towards mainstream issues at regional level where they are more viable participants in governments. Consequently, the regional level coalition bargaining can be immediately reflected in the moderation at national level elections afterwards. H9: The decentralization level of a country impacts the radicalization of the ERPs regionalist issues. As for the economic situation of the country, Muller- Rommel (1994) initially assumed that the potential for a center-periphery conflict rises in period when national resources become scarcer. Sorens (2012), on the other hand, has proved the opposite direction (at world level). He argues that in the higher GDP countries the vote share of the ERPs is higher. But, can this be generalized on the parties ideology as well? It could be expected that the radicalization is more possible in richer countries where an ERP would have more incentives to self-administer a region that it would have had in a less economically progressive country. Thus, the H10 arises. H10: The wealthier countries provide more incentives to the ERPs for radicalization of their self-determination territorial demands than the poorer ones. Finally the ERPs are known for being very EU optimistic. Their optimism and support is driven from Europe for the Regions Program in which they find justification for their territorial claims. In spite of the fact that some of the secession demands differ regarding the independence in EU or independence out of EU, they are expected to be more radical when the EU regional institutional support is present. Convergencia I Unio in Spain, has even got to suggest that a European Council of the Regions is needed in order to demonstrate their willingness to be united in EU and be directly included in the EU institutions through its region, where the Catalan language could be used in the EC institutions (Muller-Rommel, 1994). H11: The EU Membership status encourages ERP s radicalization due to the possibility of profiting from the EU Regional Policies. Method specification and Data With logistic regression statistical inference, I will assess the significance of the influence that the presumed factors have in predicting radical or secessionist ideological stances of the ERPs in Western Europe. The logit and probit models demonstrated almost no different outcomes, thus, I will only report the logit models (Demaris, 1992; Liao, 1994; Kohler and Kreuter, 2009). I measure ERPs radicalization of the self-determination demands, which is the dependent variable in this analysis, with two binary variables (twofold categorization of the parties): the first one takes the moderates vs. radical and second one the federalists vs. separatists (excluding the protectionist parties). I exclude the protectionist parties from the dataset (15% of its cases) in order to see by how much the final outcomes can differ if the parties whose self-determination demands are driven mainly from their need for cultural or linguistic identity protection are out of the game. I will report the results of the logit estimates from the maximum likelihood method for both type of dependent variable measure, and I will also compute the marginal probability effects for the independent variables for the two measures of the dependent one. First I will present tables reporting the marginal effects on the probability for radicalization when the instantaneous marginal effects for the categorical variables are taken into consideration and the dummy variables go from zero to one. Then, I will present a table of the marginal effects of the probabilities of radicalization when the independent variables change for one unit, which in the case of the dummy variables is again a zero to one change, but for the continuous variables the change goes from one standard deviation below and one standard deviation above the mean, having all other variables held at their means. The logit models cover nine Western European countries (Belgium, Denmark, Finland, France, Germany, Italy, Spain, Switzerland, and the United Kingdom) and are based on 414 cases that enclose 44 ethno-regionalist parties competing in 28 core regions. The largest proportion of almost 28% of the complete cases belongs to Spain, and the majority of cases per region belongs to Faroe Island with 16% of the total cases. The national elections take place from 1945 to In Appendix 1 a summary of the party cases and the number of core regions by country will be presented. Appendix 2 offers a summary of the number of cases per core region, and Appendix 3 summarizes the period of democratic elections that the dataset covers and the number of elections by country. Finally, to check the robustness of the findings, I present the same logit regression models by firstly excluding the country which offers most of the dataset cases (Spain) and then excluding the region offering most of the cases by region in the dataset (Faroe Island). The findings are enclosed in the Appendix 4. Page7

8 Variables measurement: The dependent variable in this empirical comparative research is the level of radicalization of the ERPs demands for self-determination in national elections campaigns. As introduced earlier in this article the dependent variable will be constructed on the basis of the Massetti and Shakel s (2013 a, b) classification of the parties scores regarding radicalization of the territorial self-determination. They have used two categorizations of regionalist parties. The first one is a dichotomous variable which codes zero if a party is moderate (autonomist) or one if it is radical (secessionist). The two categories contain other two subgroups of ideology radicalization levels: the moderates are constituted of protectionists and federalists, and the radicals can be ambiguously separatists or explicitly separatists (Appendix 1). The division between moderates and radicals is found to be robust and it can travel safely through regions and space because the basic distinction between parties that want some degree of selfgovernment and parties that want the region to break away from the state is understood in the same way in different regions or states. Moreover, the initially introduced definitions of ERPs ideological stances by Jane (2007), Muller-Rommel (1994), Dandoy (2010), Gadjanova (2013) or Massetti and Shakel (2013 1, b) all start with the lowest level of radicalization (protectionist) and end up with the highest level (secessionist). There are several reasons why I have decided to use this codification sheme before others such as: the Ethnoregionalist Index by Gadjanova (2013), The Chappel Hill Surveys by Hooghe et. al. (2010), the EPAC dataset by Szocsik and Zuber (2012) etc. Firstly, and more importantly, this codification is the most compact one in terms of time and space. In fact it covers more cases than I actually include in my analysis, which has been restricted by the definition of the ethno-regionalist parties I decided to use for the purpose of this article. This implies that the other datasets are limited in time and space, as well as the number of parties that they cover. For instance Gadjanova s Ethno regionalist Index relies on the Comparative Manifesto Project data (Volkens et.al., 2001) which has been already criticized for being subjective in the country and party cases which depend on the subjective judgment of the party manifesto s coders in deciding the importance of the parties in national elections through countries. Thus, many of the ERPs have been excluded, which leads to less covered cases by the index as well. Instead, the coding I use has been validated with the territorial model of the EPAC and proved to be in a very high correlation, thus it increases the reliability of the findings. This classification very much coincides with some of the classifications of the ERPs introduced above such as the Rommel-Muller s one (1994) or the Dandoy s one (2010). iv The final argument is due to the multilevel nature of the coding. Namely, the authors have also provided coding for the regional election party appeals, which in a further stage can be integrated in the analysis. Therefore, the first dependent variable will measure the radical versus the moderate stances of the ERPs by taking zero score when moderate and one when radical. Further on, I use the above mentioned coding scheme to construct another dichotomous dependent variable that will deal from one side with only separatists (the ambiguously secessionist and secessionist parties) which score one, and from the other side the federalist parties only, excluding the protectionist ones from the very category and the model. Appendix 5 includes the parties and the ideology codification on the two dichotomous variables. The first independent variable is the ERP s Inclusion in Government which is a dichotomous variable that indicates whether the party has been included in regional or national government at the moment of national elections. Thus, it indicates one if, before the national elections, the ERP had been included in any of the local or national governments, and zero if it has not been included neither in regional nor in national government. Due to very small number of cases where an ERP has been a coalition partner in a national government only, we cannot test this variable separately (Nordsieck, ). To test the H2, H3, H4 and H5 I will construct four variables that will indicate the issue policy stances of the two largest mainstream parties towards the two general regionalist issues presumably in the hands of the ERPs. The first is the Territorial Issue, and the second the Issue of Multiculturalism. Meguid (2005, 2008) takes the decentralization as the core issue in ERPs ownership. Gadjanova s (2013) Ethno-regionalist Index v. on the other hand is based mainly on two issues, decentralization and multiculturalism. Thus, the statewide party competition cluster of variables includes variables that will measure the positioning of the mainstream parties towards the regionalist issues and the effect they have on the radicalization of the territorial claims. For this purpose the measures of the Party Manifesto Project Data are used (Volkens et al. 2010, 13), and the variables rely on the percentages of the quasi-sentences devoted to several issues in the party manifestoes for Page8

9 national elections. The territorial issue (decentralization, under code per301 and centralization under code per302), and the multiculturalism (multiculturalism positive-per607 and multiculturalism negative-per608) will be taken into consideration. The variables should measure the salience of each of these issues for the two biggest mainstream parties. Each of the mainstream parties positions towards the regionalist issues is constructed as a variables that firstly takes the percentages of the quasi-sentences devoted to the issue, from party manifestoes for national elections, and weights them by party size for the two largest mainstream parties combined. In sum, the score indicates the salience of the issue for the two largest mainstream parties contesting the national elections. The larger the number the more salient the issue is for the two largest mainstream parties. And so, I have four variables that will indicate two opposite mainstream parties stances on the territorial and multiculturalism issues. The first one is the Decentralization Salience which shall indicate an accommodative position of the mainstream parties on the territorial issue when high, dismissal when low. The Centralization Salience indicates the adversarial position on the territorial issue when high, and dismissal when low. Multiculturalism Positive variable indicates an accommodative position on the issue by the mainstream parties and the Multiculturalism Negative indicates adversarial position of the mainstream parties on the multiculturalism issue when high as well. Therefore, the expected directions following Meguid theory s logic would be positive in case of accommodation and negative in case of adversarial. That is to say, when the two mainstream parties take a high level of accommodative position in the regionalist issues, the ERPs should behave radical in order to protect the ownership of the issues. In the case of a high level of adversarial tactics towards the regionalist issues the ERPs are expected to react moderate because they do not have to react on the mainstream provocation given that they still have the ownership of the regionalist issues and they have for granted support of their electorate. If the theory claiming that ERPs are inflexible regionalist issue players holds, the hypothesis on both issues should behave in the same direction. In case that the other direction of the hypothesis is confirmed that would mean that ERPs are not depending on the mainstream parties in creating their strategies, but they rely rather on the specificity of the issue or their own strategies that involve other parties (electoral players). ERP Competition in National Parliament is a dichotomous variable that tells whether there was another ERP from the same ethno-regional group holding seats in the National Parliament at the moment of national election. Thus a score of one indicates a presence of at least one other ERP in National Parliament before elections, and a score of zero indicates absence of the phenomenon. Party-Family Parliamentary Competition is a dichotomous variable as well and tells us whether there was another state s ERP holding seats in the National Parliament at the moment of national election. This variable differs from the previous one by the fact that it measures the regionalist competition in the national legislative bodies at national level, unlike the previous one that was speaking of the competition within the same ethno-national group. Thus a score of one indicates a presence of another ERP in the National Parliament, no matter which region it represents. ERP Electoral Competition is a dichotomous variable that measures whether other ERP(s) address the same ethnoregional group, which is considered as an intra-group regional competition. Thus, a score of one indicates that there was at least one other ERP competing within the same region for the same electorate at the given national election, and the lack of other ERP competitor is coded with zero. ERP size is a variable that is measured as a percentage of total vote share gained by an ERP at national elections. I include their total votes at national elections, not only the votes from their core regions, because in this model I assume in many cases that their votes don t necessarily come from their region and from a closedgroup of electorate, but they appeal to a larger electorate by offering wider scope of issues. PF=Parliamentary Fragmentation, is a variable that weights parties by size and it is obtained by first taking the seats share of each party as a decimal, squaring this value, and summing these values for all parties. Independents are ignored. The figure obtained is then subtracted from 1 to produce the figure for electoral fragmentation. This score times 10 gives the final score per year of election. I take the measures from Siaroff s (2000) dataset, and these figures are proxy for the proportionality of the system. Consequently, the higher the value, the more fragmented the parliament is, and the more proportional the system is. Thus, it is expected that the less disproportional the system the more opportunity for office and government inclusion, more opportunity for vote-maximizing, and so more moderate the ERPs would be. Decentralization variable is measured in a form of a RAI=Regional Authority Index (Hooghe et al, 2010) that measures the level of decentralization which distinguishes between self-rule (authority exercised by a Page9

10 regional government over those who live in the region), and shared rule authority (authority exercised by a regional government or its representatives in the country as a whole). It can range from 0 to 24. GDP per Capita is an economic indicator of the state within which the ERP competition happens. The data is taken from the World Bank database and it indicates the GDP per capita (current US $). EU Membership is a dichotomous variable that demonstrates whether the nation-state was a member of the European Union or not. With a score of one it indicates the membership status and zero if the country in the given election year was out of the EU Family. Table 1 provides a summary of all the variables and their measurements. Thus the first thing we can notice is that by taking the protectionist parties out of the dataset the models start with a difference of 63 cases (the variable excluding them counts 15% less cases). This directly influences the probability of having a radical party when only the dependent variable is observed; it increases the probability from 46% in the full dataset to 55 % in the restricted one. The variables that refer to the salience of the ethnoregionalist issues for the two largest mainstream parties demonstrate variation of the scores which indicates that they take different stances for different issues. If they range from 0 to for the Decentralization Salience Issue, they go from 0 to 2.33 in the case of centralization Salience Issues. This means that mainstream parties in general took higher scores on the Decentralization Issue with an average of 2.37 than on the Centralization one (mean of 0.19), but also with a higher standard deviation which indicates less stability across cases. The variation is lower when the Multiculturalism Issues are compared with a mean of 0.52 versus 0.45 and higher standard deviation (1.46) for the Multiculturalism Negative. The probability of finding an ERP in National Parliament from the same region is 33%, which is more than double lower when we look at the state level ERPs included in National Parliaments (73%). In 56% of the cases ERPs find a regionalist competitor at elections. The cases included in the dataset on average have 15.8 points on the RAI Index, which so far can range from A very high degree of the cases included in the analysis have been part of the European Union (78%), and the mean party size in the dataset has something above 6% electoral vote share at national elections. The maximum though, can go up to 55%. Table 1. Descriptive statistics summary of the variables N Mean Std. Dev. Min Max ID_Moderates_Radicals ID_Federalists_Separatists In Government Decentralization Salience Centralization Salience Multiculturalism + Salience Multiculturalism - Salience Regionalist Parliamentary Competition Party Family Parliamentary Competition Regionalist Electoral Competition Party Size Parliamentary Fragmentation Decentralization Level GDP per Capita *log EU Membership Page10

11 Results: Table 2 and Table 3 contain the estimates of the logistical odds and the marginal probabilities for radicalization respectively, with respect to the twofold division of the party cases, the one between moderates and radicals (for the purpose of this article called MR Set), and the one between federalists and separatists (the FS Set). The three models in each Set are based on the same independent variables and they only change the dependent variable as already introduced in the variables operationalization section. Starting with Table 2, we introduce Model 1 which includes variables that should test the government inclusion theory and the mainstream party influence on the ERP s ideological positions. Model 2 includes the intra-group party competition by adding the variables on regionalist, internal-group and party family competition. The final Model 3 includes country variables such as GDP per Capita and Decentralization Level, but it also controls for the party size and electoral system proportionality to be able to test the ex-post electoral figures and its influence on the ideological profile of the ERPs. The EU membership status is the final control variable included in the final Model 3. We can see that across all the cases we can firmly state that the theory that inclusion in regional government of an ERP moderates the ideological appeal towards selfdetermination has been confirmed. Namely, the logistic odds for the party to radicalize go from to among the three models in the MR Set, and from to in the FS Set. Regionalist Parliamentary Competition and the Regionalist Electoral Competition (in spite of the fact that the later drops from 0.01 to 0.05 significance level in both Set MR and Set FS for Model 3) firmly confirm the hypothesis. Thus, when an ERP has been included in a National Legislative Body before national elections, more radical parties can be expected to arise. On the other hand, in the cases where there is one or more ERPs competing with the ERP of our interest the chances to find this party as moderate increase. Party Family Parliamentary Competition variable (parties that might also belong to another region with salient territorial politics) goes in the expected negative direction proving moderation when other good examples of regionalist representation exist within the National Parliament. Nevertheless while the significance level is at Alpha 0.01 in the MR Set, it fails to prove significance in the Model 2 under the FS Set and it gets Alpha 0.05 significance level in the Model 3. Fragmentation of the Parliament also proves the hypothesis (at Alpha 0.01 level) that the more proportional an electoral system, the more small party size inclusive it is, and as a result the odds to find less radical parties within the ERP family are higher. While the decentralization level also proves to be negatively related with radicalization the economic performance of a country proves the opposite - the more economically developed countries are a more fruitful soil for radical parties. Even if they tend towards the expected directions Party Size (negative) and EU Membership (positive), these variables fail to prove statistical significance, i.e. the effect they have on the radicalization is not statistically different than zero, given that there exist a little multicollinearity between this and the other independent variables in the model. And finally, I will go back to the set of variables that test the mainstream parties competition relevance for the ERPs strategies. The final Model 3 proves the variables that measure the mainstream parties positions towards the salience of the decentralization and multiculturalism issues, to be insignificant. Nevertheless, Model 1 and Model 2 of both MR and FS Sets demonstrate significance at 0.01 Alpha level for the multiculturalism issues. More precisely, these models prove that the accommodation strategies of the two mainstream parties towards the support for the issue of multiculturalism provoke moderate reaction, i.e., less radical parties. Right on the contrary, when the two mainstream parties take the adversarial position and stand against multiculturalism in their party manifestos, the ERPs reaction is radical. Both directions go opposite way of the theoretical expectations on the ownership of the issue position. Despite of the fact that the decentralization and centralization issues have not proven to be significant, we can note that in the case of testing the territorial issues the direction supports the logic of Meguid s theory. Page11