Types of Economic Voting in Regional Elections: The 2012 Catalan Election as a Motivating Case

Abstract

The standard assumption that economic voting (EV) is “jurisdiction-specific” inevitably leads to a breakdown between “national EV” and “regional EV”. This paper challenges this overly simplistic distinction by proposing a more complex typology, whereby national and regional incumbents may be assessed in both national and regional elections, according to either national or regional economic conditions. Accordingly, new and more sophisticated types of EV emerge, such as “second-order EV” or “coattail EV”. In this paper, some of these new types of EV are verified with a suitable case study. The 2012 Catalan election was carried out in the context of severe recession, but also under the impression – among many Catalans – that the economic policies of the Spanish government were harshly punishing Catalan economic interests. Binomial logistic regression models confirm that, under political circumstances such as these, voters may use regional elections to assess the national incumbents' economic performance, whereas regional incumbents may end up exonerated from poor economic performance. This case study may be illustrative for other regional elections around the world.

ORCID

Agusti Bosch http://orcid.org/0000-0003-3972-9619

Acknowledgements

Previous versions of this article were presented at the XI AECPA Conference, the 7th ECPR General Conference, the UAB Ph.D. Seminar, the UPF Political Behaviour Seminar and the Rosa Virós Forum. I want to thank the valuable feedback received by the colleagues there, as well as the comments from Michael Lewis-Beck, Guillem Rico, Lluís Orriols, Jordi Muñoz, Nacho Lago and the three anonymous referees. I also want to thank the data providers: ICPS, MEDW and CIS. I am grateful for the willingness of the ICPS research team (especially Lucía Medina) to include in the questionnaire a question on attribution of responsibility, which was used in this study.

Disclosure Statement

No potential conflict of interest was reported by the author.

Supplementary Material

Supplemental data for this article can be accessed at 10.1080/17457289.2015.1119153

Notes

1. The two types are those represented in cells ‘c’ and ‘d’ in Figure 1. There are no accepted denominations for those types of EV, hence the names c-type EV and d-type EV.

2. Notice Figure 2 renames REV and NEV to distinguish them from these contaminated types of EV.

3. Notice their concept of second-order EV is somewhat different from ours because their characterization is based on voter's assessment of local economic conditions, whereas ours is based on national ones. In fact, their second-order EV is more similar to our ‘coattail EV’.

4. In the whole text, the term ‘national’ refers to internationally-recognized nation-States such as Spain and the term ‘regional’ refers to communities – such as Catalonia – that are not independent nation-States. We know most Catalan natives would conversely use the term ‘nation’ to refer to Catalonia and the term ‘state’ to refer to Spain. However, our terminology aims merely at simplifying the reading for an international audience and is in no way meant to deny the status of ‘nation’ to Catalonia nor is it meant to have any political implication.

5. Curiously, the regression coefficient for the newly introduced Spanish economic situation attains some statistical significance but the sign of the coefficient is in the opposite direction than expected, that is, those who evaluate the Spanish economic situation as more positive, have a lower tendency to vote for the Catalan incumbents. This is simply because they have a higher tendency to vote for the Spanish incumbents, as panels 3.3 and 3.4 will confirm. Thus, the results for the genuine REV model remain negative. It must be also acknowledged that the coefficient instability could be the result of multicollinearity (see diagnostics in the table note). This danger advises to give a less prominent role to the models with both perceptions, just as controls of the main models.

6. Admittedly, the coefficient for this variable fluctuates and it even loses some significance due to the high correlation among the two perceptions (r = 0.77). However, its significance level is still impressive (p = .013). Taking into account this was a very demanding test (because we know that collinearity frequently weakens the significance levels), the results indicate a reasonably robust performance of this coefficient.

7. Notice that the argument is not that second-order EV is a novel behaviour in Catalan elections: as Table OA7 (in the online appendix) shows, such a model always worked in past elections. Instead, the novelty is that only this model continued to work while genuine REV ceased.