Research

Working Papers:

• Political Competition When Gangs Rule: The Effects of Removing Armed Groups' Territorial Control in Brazil  (Job Market Paper)
  
  Abstract: Millions of citizens of democratic countries live in areas captured by non-state armed groups. In this paper, I study how territorial control by criminal organizations affects political competition and public service provision. I exploit the staggered implementation of a pacification policy in Rio de Janeiro that restored state control in shantytowns ruled by drug gangs using difference-in-differences. I show that removing criminal rule increased electoral competition in pacified slums. After pacification, votes became more dispersed across candidates for city council, the party in power lost vote share, and candidates running for the first time performed better. To study the effect of pacification on public service provision, I constructed a novel dataset of public services that city councilors propose. I show that city councilors request more public services to pacified territories, but only those where they have a political base.

• When Democracies Refuse to Die: Evaluating a Training Program for New Politicians (Draft coming soon)
  Joint work with Ernesto Dal Bó, Claudio Ferraz, and Frederico Finan.
  
  Abstract: Many countries have experienced democratic backsliding in recent years, a process often blamed on rising distrust in the political class. In response, non-profit organizations have created programs to recruit and train a new class of political leaders. But whether this response by civil society can succeed in renewing the political class remains understudied. This paper evaluates Brazil's largest non-profit, non-partisan training program for aspiring politicians. Based on knowledge about the selection process, we leverage quasi-random variation in the probability of being accepted to the program using a regression discontinuity design to estimate the programs’ effects. The training program succeeded in motivating aspiring politicians to run for office. Trained aspirants are 70 percent more likely to run for office than the control group. Conditional on running, trained candidates raise more money and run more competitive campaigns. However, we do not find evidence that they are more likely to be elected. While the training program helps to renew the political class by inducing political entry, newcomers are still unlikely to be elected. This limits the effect of the training program on political selection in the short run. 

Work in Progress:

• The Expansion of Drug Syndicates and Homicides in Brazil
  Joint work with Mariana Carvalho and Vinicius Peçanha

• "Coronavirus is finished": Beliefs and Behaviors in the Tanzanian Misinformation Bubble
  Joing work with Gabriel Brown and Munir Squires

Publications:

• Property-Level Assessment of Changes in Forest Clearing Patterns: The Need for Tailoring Policy in the Amazon. Land Use Policy, 2017.
  Joint work with Juliano Assunção, Clarissa Gandour, and Romero Rocha.