Brazil has witnessed the rise of a new political phenomenon: there is an increasing number of candidates with a background in the security forces who leverage their professional experience to signal commitment to tackling public insecurity, often through punitive and repressive approaches. In this paper, I identify them as law-and-order and examine the effects of violence occurring shortly before elections on support for their candidacies. Drawing on the issue ownership framework, I argue that a positive relationship is expected in areas plagued by high levels of crime, as exposure to violence during the electoral window heightens the salience of tough-on-crime platforms. I provide suggestive evidence in support of this claim. Using a dataset containing the coordinates and dates of gunshot incidents in Greater Rio de Janeiro and Recife, and leveraging the as-if-random timing of violence, I implement a regression discontinuity in time design. I compare the share of votes for law-and-order candidates in polling stations that experienced a nearby gunshot incident shortly before the 2018 general election to those that had incidents shortly after. While findings point toward a potential causal link, especially in Recife Metropolitan Area, the results are sensitive to the criteria used to classify polling stations as treated and control. That is, treatment effects' significance varies depending on how spatial and temporal proximity of gunshot incidents is defined.
Victor Lopes (Fri,) studied this question.