Electoral Campaign Attacks: Theory and Evidence

• In Event: Models of Electoral Competition

Abstract

This paper studies the main determinants of electoral campaign attacks, both theoretically and empirically. We first propose a model of electoral contests, where the probability of winning an election depends on the candidates' initial levels of political support plus their binary (simultaneous) decisions to attack or not their opponents. A campaign attack is assumed to be costly and to cause a fraction of the targeted candidate's support to switch to her opponents. We provide a complete characterization of the equilibrium in races with two and three candidates. Under single ballot plurality, we show that the 2nd place candidate is always the most aggressive candidate. Moreover, candidates with an electoral advantage are always comparatively more likely to receive an attack.

Next, we leverage the tractability of our model to study two comparative static questions to better understand how the incentives to attack depend on race characteristics and voting rules. First, we show that, in three-candidate races, the two front-runners become less aggressive towards each other when the support for the 3rd place candidate increases. Intuitively, a campaign attack benefits not only the attacking candidate but also other opponents of the targeted candidate. Our analysis shows that the top two candidates are particularly concerned about this spillover (``dillution") effect when the 3rd place candidate is strong. Next, we show that the pattern of campaign attacks differs in significant ways under single and dual ballot plurality, reflecting the distinct nature of electoral competition under both systems. Specifically, focusing on the behavior of candidates in the first round of elections, we show that the 3rd place candidate is always the most aggressive under dual ballot plurality.

Empirical studies on campaign strategies are usually made difficult by the lack of available information on how campaigns are actually carried out. To overcome these difficulties, we collect detailed information contained in ``right of reply" lawsuits filed in Brazil during the 2012 and 2016 municipal elections. The Brazilian electoral legislation protects candidates, during the campaign period, against slanderous, defamatory, and false accusations, granting the victim of an attack the right to respond to the offense at the offender's cost. ``Right of reply" lawsuits are simple, inexpensive, and must be decided by the local electoral judge within 72 hours. Based on information retrieved from these lawsuits, we construct a unique dataset of 69,252 ordered pairs of candidates containing the precise directions of campaign attacks in 10,461 distinct electoral races.

We begin our empirical analysis by examining the general patterns of campaign attacks in Brazilian municipal elections. Focusing on electoral races with two and three effective candidates held under single ballot plurality, and controlling for a rich set of candidates' characteristics and both municipality and election-year fixed effects, we find a pattern of attacks that closely matches our theoretical predictions. Specifically, we show that the 2nd place candidate is always the most aggressive. Moreover, in three-candidate races, we find that candidates are always more likely to target their highest-ranked opponent, with the front-runner being the most attacked. Finally, we show that an increase in the electoral strength of the 3rd place candidate significantly reduces the likelihood of an attack between the two front-runners.

Next, we exploit two different research designs to investigate how specific aspects of the political and institutional environment affect campaign attacks. A robust prediction of our model is that candidates with an electoral advantage are more likely to receive an attack. To test this hypothesis, we leverage quasi-experimental variation in electoral support arising from virtual ties between 2nd and 3rd place candidates. Our approach follows Anagol and Fujiwara (2016) who used data from elections in Brazil, India, and Canada to show that close runners-up are substantially more likely to run in and win subsequent elections. We exploit a similar regression discontinuity design to show that close runners-up are about 2 percentage points more likely to receive a campaign attack in the next elections compared to close 3rd place candidates, which corresponds to a striking 160% increase. Finally, we investigate whether the pattern of attacks differs under single and dual ballot plurality. To do so, we exploit quasi-experimental variation arising from the fact that in Brazil municipalities with fewer than 200,000 registered voters must use single ballot plurality, while those above this threshold must use dual ballot plurality. Consistent with the predictions of our model, we find that 3rd place candidates become significantly more aggressive under dual ballot plurality.

Current version: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4634314

Author

• Marcos Yamada Nakaguma, Sao Paulo School of Economics, FGV