Beyond Identities: Political Determinants of Support for Decentralization in Contemporary Spain

Abstract

Support for decentralization has generally been considered an outcome derived from the existence of regional or subnational identities. A recent complementary explanation has highlighted the role of politics to create regionally orientated support. We combine these two explanations by studying what determines the support for decentralization in Spain, an almost unique case where these two approaches can be tested. Using a hierarchical model, we show that in those regions without a different subnational identity, some citizens develop regionally orientated support because of their wish to bring the administration closer to the citizens and because their preferred party is pushing for further decentralization. In addition, we highlight the importance of considering decentralization as an outcome derived from party dynamics, which are able to shape political attitudes towards decentralization.

Keywords:

• Decentralization
• hierarchical model
• regional identification
• political competition

Notes

¹ These two arguments are outside this research. They are, however, important to show that decentralization has been studied mainly taking an aggregate approach and as an elite-driven process.

² Catalonia, the Basque Country and Galicia are the so-called historical nationalities (Requejo, 2001; Moreno, 2002). Even though Andalusia gained fast track access to autonomy, this AC is not considered a historical region due to its lack of a former regional government and sociolinguistic distinctiveness (Moreno, 2002).

³ CIS 2829 (2010). n = 10 409. Includes data for all the 17 ACs and the two African autonomous cities of Ceuta and Melilla. It is freely available at http://www.cis.es.

⁴ Even though the dual measure of identity employed in this article has been criticized by the literature, especially concerning the appropriate identification of nested identities (Kalin and Berry, 1994; Goyder, 2003), this is still the most popular way to deal with identities and the only one available in the survey employed in this article.

⁵ More specifically, the wording of these two variables is: ‘How would you assess what the regional government has done in the last year?’; and ‘According to your personal experience, would you say that public services work better or worse than services managed by the State?’. Both variables are uncorrelated, thus avoiding colinearity.

⁶ We consider supporters of the PP at the regional level instead of at the national level since it is in the regional arenas that regionally orientated preferences are expressed. Furthermore, since the electoral system is more permissive in the regional arena, voters cast their ballot less strategically. All in all, results do not vary when considering conservative party voters at the national level since both variables are highly correlated (r = 0.867, p < 0.01).

⁷ All citizens that voted for any other party rather than the PP are grouped together, as the percentage of those who voted for a different option than the Socialists is substantially small. Even if we turn vote choice into a categorical variable, our results still hold. Indeed, among those that did not vote for the PP, 68.7% declared having voted for the PSOE. Only in Catalonia and the Basque Country was this percentage substantially lower (45.1% and 31.9%, respectively).

⁸ In Spain, the central government must transfer competences to regional governments, even if the Constitution grants other powers to the regions. The number of competences transferred indicates, therefore, how much power each region formally has. We take into account the devolvement of responsibilities and resources from the central government to regional governments approved until 2010 (year of the Barometer Survey). The data come from the Spanish Ministry of Territorial Policy and are available at http://www.mpt.gob.es/areas/politica_autonomica/traspasos/ (accessed 14 Jan 2013). The Basque Country, Navarre and Catalonia are the ACs that hold higher levels of self-government. However, data indicate that the self-government of the first two regions is slightly less than the one in Catalonia and even other regions. This appears to be counterintuitive and, in fact, it is due to the existence of a different fiscal system in these regions. The Spanish Constitution grants the Basque Country and Navarre some fiscal privileges from the past, mainly the capacity to collect taxes. Therefore, these competences have not been transferred, because they were already in their hands when the Spanish territorial model started to develop. To solve this discrepancy, we attribute to both ACs the same number of competences as Catalonia (187), the AC which, in 2010, had the higher number of transfers approved. Through this transformation, we manage to better fit the database with the real situation of Spanish devolution. In any case, in order to check the robustness of our results, we have re-run the model using instead a dummy variable for the Basque Country and Navarre. The same results are obtained.

⁹ Competences act as a proxy for the role of political parties. Other measures could, perhaps, have better addressed this need, such as the Regional Authority Index (RAI) database (Hooghe et al., 2010). Unfortunately, there are two reasons that discourage the use of this measure. First, these data cover only the 1980–2006 period, before decentralization reforms were carried out in Spain; and, second, there is almost no variation across ACs—decentralization always takes the value 14.5, except for Ceuta and Melilla (15) and the Basque Country and Navarre (15.5).

¹⁰ It is noteworthy to mention that there is a heated debate in Spain regarding so-called ‘inter-territorial solidarity’. In Spain, most taxes are collected at the national level and these are latterly transferred to the AC. In 2008, the Spanish government published the fiscal balances between territories. Whereas some regions received more money back than their citizens had contributed to the ‘common box’ (such as Extremadura or Andalusia), others received less money (such as the Balearic Islands, Madrid or Catalonia). These figures encouraged another period of intensive claims against this situation and, since then, wealthier regions have repeatedly called for a change in the financial system—especially Catalonia.

¹¹ Data available from the National Statistics Institute (http://www.ine.es). The method to calculate the tax scales is the ‘benefit tax incidence’ (López-Casasnovas, 2010). For further details on the calculus of the indicator in Spain, see the final report from the Grupo de Trabajo sobre metodología de cálculo de las balanzas fiscales [Working Group on calculus methodology for the fiscal imbalance] (Ruiz-Huerta and Salinas, 2006).

¹² Only in three regions (Catalonia, the Basque Country and the Canary Islands) is the percentage of vote to NSWP larger than 30%. Taking the arguments from Van Houten (2003) about the effects of the dynamics of regional party competition, it could be argued that the percentage of votes to NSWP at a given AC may put pressure on the regional branch of the socialist party, thus leading to higher support for decentralization among socialist voters in these regions. We have discarded this hypothesis since primary tests do not support such evidence.

¹³ A significant part of the variance of the intercepts and the independent variables of interest indicate that part of the variability of the dependent variable can be attributed to the second level, i.e. to the ACs. However, because the number of groups is very small (17 Autonomous Communities), it is not statistically appropriate to fit a random intercepts/random slopes model (Kam and Franzese Jr., 2007).

¹⁴ In terms of model fit, the Aikake's Information Criterion (AIC) and Bayesian Information Criterion (BIC) indexes indicate that the last model fits the data better. The intraclass correlation coefficient (ICC) indicates that a part of the variance of the intercept can be attributed to the second level. It implies that the proportion of the preferred degree of decentralization variability is due to occur between regions rather than within them.

¹⁵ An interaction between vote and congruence in government composition fails to be significant.