Principal Investigators: Taylor Boas, F. Daniel Hidalgo
Country: Brazil

Registration: 20151118AA

Dates of Intervention: September – October 2016

Background: The northeast region of Brazil is a comparatively poor area with a long history of governments that fail to serve the public good, thanks to oligarchic politics, clientelism, and endemic corruption. This project is designed to test whether information about incumbent performance can induce accountability in the state of Pernambuco, one of the most populous in the northeast region. Interventions will take place surrounding the October 2016 municipal elections, focusing on incumbent mayors running for reelection. The research will be conducted in partnership with the State Accounts Court of Pernambuco (Tribunal de Contas do Estado de Pernambuco, TCE-PE), a well-respected government institution charged with auditing and publicizing information about the performance of municipal governments.

Research Design: The experiment will be a multi-arm study. In the common arm, respondents in the treatment condition will be informed as to whether the TCE-PE’s audit found evidence of mayoral corruption during an earlier year in his/her term. We plan to sample an equal number of “corrupt” and “non-corrupt” municipalities in which incumbents are running for reelection. Alternative arms will convey information about other municipal performance indicators that ought to be more salient, particular for certain categories of voters who receive benefits. Possible performance indicators (the final list is subject to change after pre-testing) include a) successful administration of the popular, federally-funded conditional cash transfer program Bolsa Família; b) implementation of a federally-funded Family Health Program that pays for doctors and nurses to conduct house calls; c) change over time in the rate of students falling behind in school. Each of these policy areas relies on mayors and their appointed officials to administer programs that are funded by higher levels of government, and there is wide variation in the success or failure of their efforts. In all treatment conditions, municipal performance will be compared to statewide averages or, potentially, to that of a randomly-chosen neighboring municipality. Enumerators will deliver information orally and will also hand out scorecards that convey the information graphically alongside the incumbent mayor’s official photograph. The experiment will be conducted in the context of a two-wave panel study. The pre-election wave will measure baseline variables and administer information treatments. The post-election wave will measure outcome variables, including self-reported turnout and vote choice. 

Hypotheses:
  • Can providing information about government compliance with the law  change voting behavior?
  • Does providing information about highly salient policies change voter behavior more than general compliance information?