Is a corrupt government totally unacceptable?

Abstract

Corrupt governments are not always punished by voters. Under certain circumstances citizens consider voting for the incumbent party even if the party is perceived as corrupt. Using survey data for Spain, this article analyses what makes citizens reject (or not) the idea of voting for a corrupt party. Previous research has shown that party identification, ideology and political information play a role in voters’ reactions to corruption. The article argues that voters judge corruption in relative terms; what matters is not how corrupt the incumbent party is perceived to be but whether it is deemed to be more corrupt than the other parties.

Keywords:

• Corruption
• electoral behaviour
• government
• ideology
• parties

Acknowledgements

We would like to thank the two reviewers of this article for their useful comments. We would also like to acknowledge the Spanish Ministry of Economy and Competitiveness for its financial support through the project CSO2013-47071-R (PI: Mariano Torcal) and those behind the panel survey CIUPANEL 2016, used in this article.

Notes

1. See Taber and Lodge (2006) for a full elaboration of the argument.

2. Ecker et al. (2015) analyse the level of corruption in the country, but do not consider perceived corruption in the government and the other parties.

3. The survey was part of a research project financed by the Spanish Ministry of Economy and Competitiveness (CS02013-47071-R), and was directed by Mariano Torcal, Universitat Pompeu Fabra, Barcelona. The survey, conducted by NetQuest, is also part of the European Election Study, directed by Hermann Schmitt, Universities of Mannheim and Manchester. The data is weighted by vote recall.

4. At the time of the fieldwork (June 2015) the new parties Podemos (We Can) and Ciudadanos (Citizens) were not represented in the national Chamber.

5. The Pearson correlation between the independent variables are Distance*Corrup.Gov = 0.49; Distance*Corrup.Parl = 0.29; Corrup.Gov*Corrup.Parl = 0.73). However, multicollinearity has been tested resulting in the Variance Inflation Factor (VIF) of 1.88, which indicates that multicollinearity is not a problem in our model.

6. The logit coefficients can be compared since the three central independent variables are on the same 0 to 10 scale.