Let’s party! The impact of local festivities on the incumbent’s electoral support

ABSTRACT

Do local festivities affect an incumbent’s re-election prospects? Despite the relationship between local public policies and political behaviour is a central topic in political science, the effect of the organisation of leisure events on voting patterns has been largely neglected. Using data from Spain, we show that Mayors doubling the per capita amount of money devoted to local festivities during the last year of the legislature benefit by around two per cent points in upcoming elections. Two mechanisms account for this relationship. First, changes in the budget given over to festivities enhance the incumbent’s support when the spending on local festivities during the previous years of the mandate was generous. Second, local festivities bring about positive rewards when the financial situation of the municipality is stable. Overall, our article sheds light on the need to consider other domains beyond “core” public policies when assessing the dynamics behind an incumbent’s re-election.

KEYWORDS:

• Local festivities
• local elections
• incumbent
• Spain
• bullfighting

Acknowledgement

We are grateful to Ignacio Lago, Eva Anduiza, Martin V. Larssen, Wiebke M. Junk, Lasse Aaskoven, Benjamin Egerod, Daniel Bochsler, two anonymous reviewers, and to all participants in the 2017 “Jornada de Comportament Polític i Opinió Pública de Catalunya” and in the “Make Democracy Great Again” Seminar series 2017 from the Political Science Department of the University of Copenhagen, for their comments and suggestions. We also want to thank José Manuel Abad, Javier Galán and David Alameda, from El País, for sharing with us their database on bull festivities in Spain.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplementary data for this article can be accessed here. The replication materials can be found at Harvard Dataverse here: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi%3A10.7910%2FDVN%2FENB8PE.

Notes

1. Since 2010, it has been possible to obtain publicly available and detailed data on the allocation of resources within the yearly local budget. The municipalities from the Basque country and Navarre regions are the exception to this rule, as long as they do not belong to the so-called Communities of common regime, and are hence excluded from the dataset.

2. Appendix S2 in the supporting information document explores the correlates of missing values. Results show that the main factor driving the inclusion or not of the municipality in the sample is the log of the population, with larger municipalities being much more likely to be included than small ones. To circumvent this, Table S3d in the supporting information replicates the results from Model 3 in Table 1 by dropping large municipalities from the selected sample until we get a balanced panel in terms of population (column 2) and in terms of the log of the population. Results confirm the evidence found in our main models.

3. We focus our attention on the party whose mayor is in office at the end of the 2011–2015 legislature instead of the elected mayor after the 2011 elections because the mayor may have changed during the legislature, as a consequence of a vote of no confidence. The source of the data is the Spanish Ministry of the Interior.

4. Although other parties hold a larger number of mayors than Izquierda Unida (for instance, the regionalist Partido Aragonés Regionalista), they mostly govern in small towns for which we do not have data on festivities. We group them together in the ‘Others’ category.

5. It is beyond the scope of this article to explain why the PP (and to a lesser extent CiU) lost a substantial share of the votes. However, two main factors can be mentioned. First, the emergence of new political competitors in 2015, both on the left (Podemos) and on the right (Ciudadanos); second, the PP (and CiU in Catalonia) had been in charge of most of the municipalities in the 2007–11 legislature and voters may have penalised the parties for the bad economic outcomes experienced during this period (Rodon and Hierro 2016).

6. The variable takes a value close to −1 when a municipality has reduced the amount of money allocated to local festivities from a certain sum to (almost) nothing. The value for increases in the sum of money allocated to local festivities is not bounded and, therefore, in order to obtain a normal distribution and to be able to properly estimate the effect for the most part of the sample, we have attributed a value of 1 to all those municipalities that had a higher value (4.2% of the sample – these are those municipalities that more than duplicated the sum of money devoted to festivities). This criterion not only allows us having a symmetrical and bounded measure, but theoretically also accounts for the positive but marginally decreasing impact of suddenly increasing the amount of money allocated to local festivities. Note that results hold if municipalities above the value of one are dropped from the sample or if we allow the measure to have a higher upper bound – yet, the one-point impact of the percentage change in the per capita amount of money spent on festivities between the 2011–2013 period and 2014 becomes smaller as the observed pattern is not monotonic but rather asymptotic.

7. Appendix S2 in the supporting information document shows that, once controlling for other covariates, the per capita amount of money devoted to local festivities is higher near election periods (2011 and 2014). The inclusion or exclusion of the control variable that captures the mean per capita amount of money spent on festivities between 2011 and 2013 does not change our results.

8. Quantile regression estimates the conditional median (or other quantiles) of the response variable. The use of quantile regression is appropriate when the conditions of linear regression are not met, in most cases due to the presence of outliers in the response measurements (see Verardi and Croux 2009). As a robustness check, table S3a in the supporting information document replicates the analysis by using a robust OLS regression. This estimation procedure enables us to prevent results to be highly sensitive to outliers and, thus, compromising the validity of the regression model.

9. Data comes from the newspaper El País and does not contain information from the Autonomous Community of Castilla-La Mancha. This region has 919 municipalities, 155 of which were included in the original festivals database, thus reducing the model from 2,056 observations to 1,901. If we use alternative specifications of the variable (e.g. a count of the number of bull events taking place in the municipality during the year, or its logarithmic transformation) we obtain the same results.

10. Table S3b in the supporting information document shows the different results of the interaction models, using a series of quantile regression models with region fixed effects. Table S3c does the same by using a robust OLS regression.

Additional information

Notes on contributors

Marc Guinjoan

Marc Guinjoan is a postdoctoral researcher at the Institutions and Political Economy research Group (IPErG) of the Universitat de Barcelona. His research interests include the study of institutions, political behaviour, elections and populist attitudes, as well as identities and nationalism. www.marcguinjoan.cat

Toni Rodon

Toni Rodon is an assistant professor at the Department of Political and Social Sciences of the Universitat Pompeu Fabra (UPF). His research interests include electoral participation, political geography, comparative politics and historical political economy, as well as public opinion and the study of nationalism. www.tonirodon.cat