When the partisan becomes personal: Mayoral Incumbency Effects in Buenos Aires, 1983–2019

ABSTRACT

A burgeoning literature finds that incumbency effects reflect mostly a personal rather than a partisan advantage. We attribute this to incumbents’ mobilization incentives. Incumbents have weaker incentives to exert costly effort on behalf of their copartisans in national races than in local ones, where their local power is at stake. We examine these implications in the Province of Buenos Aires, Argentina’s largest subnational unit, where midterm elections give mayors a strong incentive to help their copartisans running for the local council, but much weaker ones to support those running for a national seat. Using a regression discontinuity design, we find a large positive effect of incumbency in local mayoral and midterm elections. In contrast, local incumbents neither help nor hurt their copartisans running for the presidency or the national legislature.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

2 In 1983, the PBA had 125 municipalities. Since then, a series of splits (mostly in 1995) increased the number to 135.

3 Figure A4 shows an almost identical pattern in federal elections.

4 This is scheduled to change in 2027, when mayors elected in 2019 will not be eligible to run for a third term.

5 See Table A2 in the Appendix. We do not have the names of mayoral candidates for previous years.

6 That is, voters cannot pick a mayor from one party and a list of councilors from another. In 1983 seats were distributed using the d’Hondt formula with no threshold; afterwards, the Hare formula with a large threshold (one Hare quota) was employed. After remainders have been allocated, all surplus seats, if any, go to the most voted party.

7 In 2003, the presidential election was held in April, whereas elections for all other offices took place in September.

8 Registration is automatic for all citizens over 18 (16 since 2013), and mandatory voting (for citizens between 18 and 70) means that turnout is pretty high – between 1993 and 2019, average turnout at the municipal level was 80.9%.

9 The order of the candidacies is as follows: president, national senators, national deputies, provincial governor, provincial senators, provincial deputies, and mayor-councilors.

10 “Instructivo para Votar Cortando Boleta,” available on YouTube.

11 “Un Intendente Massista Reparte Tijeras para Impulsar el Corte de Boleta,” La Nación, 7-OCT-2017.

12 The establishment of open primaries in 2011 introduced another source of intra-party friction, as candidates who are defeated in the primary may end up supporting other parties in the general election (Clerici, Cruz, and Goyburu 2020). This further reinforces our point that intra-party collaboration cannot be taken for granted.

13 “The Glass Empties for the Kirchners,” The Economist, 20-JUN-2009.

14 “Datos de las 2.15. Dura Derrota de Kirchner,” La Nación, 29-JUN-2009.

15 “El Peronismo Llega Dividido y Cargado de Tensiones,” La Nación, 14-AUG-2011.

16 “Cortocircuitos en el Plan de Campaña del Kirchnerismo,” La Nación, 9-AUG-2011.

17 “El Corte de Boleta Perjudicó a Oficialistas y a Opositores,” La Nación, 27-OCT-2015.

18 See, e.g., Leandro Pérez, “Elecciones 2019: Cortes de Boleta y Cinco Batallas por el Poder Territorial en el Gran Buenos Aires,” Clarín, 26-OCT-2019; and Maías Russo Coroman, “Los Intendentes de Juntos por el Cambio Hicieron Valer el Corte de Boleta y Retuvieron Importantes Distritos del Conurbano,” Infobae, 28-OCT-2019.

19 The former is from the Junta Electoral de la Provincia de Buenos Aires. Data on national elections is from Lupu and Stokes (2009) for 1983-2003, and R’s polAr package (Ruiz Nicolini 2020) for 2005-2019. The latter only reports preliminary results. In practice, preliminary and definitive results rarely differ by more than a percentage point. Note that this does not affect our running variable because we only use this data to construct outcome variables.

20 For seat allocation purposes, mirror lists are treated as a single list. We thus add up their votes.

21 The official PJ lists in 1985, 2005 and 2017 are thus the Partido Renovador, the Frente Para la Victoria and Unidad Ciudadana, respectively. The UCR is much less problematic. We coded it as part of the Alianza and Cambiemos/Juntos por el Cambio coalitions in 1997–1999 and 2015–2019, respectively, and as a single party otherwise. We treat the “radicales K” phenomenon in 2007 as a split against the official party leadership, and thus ignore it.

22 The third largest “party” after them are “vecinalista” parties, i.e. purely local forces. These are important in some municipalities, but triumphed in just 35 elections overall.

23 Concurrent years are 1983, 1987, 1991, 1995, 1999, 2003, 2007, 2011, 2015 and 2019. We ignore the handful of mayoral elections held in midterm years, where a mayor was elected to complete an unfinished term.

24 We thus restrict the sample to incumbents elected between 1983 and 2015.

25 RD designs estimate a local average treatment effect, or LATE, for observations at the RD cutoff.

Additional information

Funding

This work was supported by Asociación Mexicana de Cultura A.C.