research

articles

How criminal governance undermines elections [paper] [press: Pindograma, Piauí] Conditional Accept at the American Journal of Political Science

  • Recipient of the Franklin L. Burdette/Pi Sigma Alpha Best Paper Award, APSA 2022
  • Recipient of the Best Paper Award from the Conflict Processes Section, APSA 2022
  • Recipient of the Elinor and Vincent Ostrom Prize for Best Graduate Student Paper and Presentation, Public Choice Society 2022
Abstract Existing theories of criminal governance indicate that criminal groups can influence elections, sometimes acting as brokers. I argue that criminal groups’ capacities as brokers stem from their control over territory and, subsequently, over voters. I theorize two criminal brokerage strategies, both made possible by territorial control: (1) gatekeeping limits rival candidates from entering the criminally governed area and (2) corralling influences vote choice. I test my theory in Rio de Janeiro, Brazil, using a natural experiment that leverages exogenous variation in voter assignment to voting booths, a novel dataset on criminal involvement in elections, and qualitative interviews. I show that fewer total candidates receive votes in areas with high gatekeeping, while criminally-aligned candidates are more likely to win in areas with high corralling. Both gatekeeping and corralling are associated with a decrease in electoral competition. These findings bring the literatures on clientelism and criminal governance closer together.


Limiting aggressive policing can reduce police and civilian violence
World Development, 2022: 105961, p. 1-18. [paper] [policy writing: Washington Post] [press: The World, Piauí, O Dia, ADPF Brazilian Supreme Court Proceedings]

Abstract Governments in the Americas rely on aggressive policing tactics to fight crime, despite scant evidence of impact. While recent studies depict militarized policing as a driver of violence, few governments have reconsidered their use of it. What impact does a restriction on aggressive policing have on violence, and why? This paper examines limits on police use of force and how they can be implemented to reduce both police and civilian violence. I argue that reforms that require internal, non-police oversight can be effective institutional constraints, minimizing police violence. In settings where organized crime is widespread, these limits can have spillover effects and further decrease civilian violence by (1) slowing the territorial diffusion of criminal conflict and (2) making conflict more predictable. I test these claims by examining an abrupt limit on police raids in Rio de Janeiro, Brazil. I find that limiting raids – militarized police strikes targeting criminal gangs and communities under their control – led to a 66% decrease in police killings and a 58% decrease in homicides. The effects were concentrated in police precincts where rival criminal groups are in close proximity. Limiting raids did not lead police to be more violent during ordinary patrolling duties, and did not affect property crimes. The implication is that restraining police use of force in high-violence settings may save lives and be no worse than hard-on-crime strategies.


How do Covid-19 stay-at-home restrictions affect crime? Evidence from Rio de Janeiro, Brazil
EconomiA, 2022: 147-163. (with Ana Paula Pellegrino). [paper] [policy writing: Nexo]

Abstract How do changes in mobility impact crime? Using police precinct-level daily crime statistics and shootings data from the state of Rio de Janeiro, Brazil, we estimate that extortion, theft, and robberies decrease by at least 41.6% following COVID-19 mandated stay-at-home orders and changes in mobility in March 2020. Conversely, we find no change in violent crimes, despite fewer people being on the streets. To address the relationship between crime and mobility, we use cellphone data and split the precincts into subgroups by pre-Covid-19-related restrictions mobility quintiles. We estimate a similar average decrease in extortion regardless of a precinct’s previous activity level, but find that the decrease in theft and robberies is substantially higher for the more mobile precincts while it disappears for the least mobile precincts. Using daily cellphone mobility data aggregated at the police precinct level, we find that changes in mobility while the stay-at-home order is in place only have a meaningful effect on robberies, which increase in likelihood when a precinct’s mobility ranking is higher than the previous day. Together, these results suggest that the stay-at-home order and associated decline in mobility strongly affected extortion and property crimes while not interfering with the dynamics of violent crime. These findings support the hypothesis that violent and property crime follow different dynamics, particularly where there is a bigger impact of organized criminal groups.


book project

Machine gun politics: why politicians cooperate with organized crime

An estimated 13% of Latin Americans live under criminal governance. Across the region, criminal groups influence minute details of civic life, including electoral politics. Residents of criminally-controlled areas constitute a large share of the electorate and, if voting in bloc, could single-handedly sway electoral outcomes. Intermediaries with such a high degree of influence or control could be appealing partners to candidates looking for an electoral edge. Criminal groups’ influence over voter behavior drives three questions underpinning my research for this book:

  • Why do candidates strike electoral bargains with criminal groups?
  • Under what conditions is cooperation with criminal groups a winning electoral strategy?
  • And what are the implications of such partnerships for residents of criminally-controlled areas?
Short description There is broad consensus that criminal groups can influence elections. However, much of the literature focuses on criminal groups' incentives to engage in politics, leaving candidates' incentives in a black box. This book contributes to our understanding of the relationship between organized crime and politics by shifting the focus to politician incentives and behavior in a study of the case of Rio de Janeiro, Brazil.

Empirically, electoral cooperation between candidates and criminal actors is common, yet it is underexplored in the literature relative to more violent or coercive interactions. That candidates would strike collusive deals with organized crime could seem puzzling: partnering with criminal groups could be toxic to a candidate's reputation, criminal groups are known perpetrators of political violence, and it is often illegal to associate with them. Existing scholarship reflects these risks, often depicting candidates as one of three limited archetypes: (1) as pawns, only engaging with organized crime under duress, and having little agency; (2) as criminals, where the candidates mitigate such risks by becoming deeply embedded within the criminal organization (or are often fellow members or leaders of the criminal group); or (3) as partisans, where political parties may have institutional incentives to strike deals with criminal actors, and candidates act as executing agents of their party.

However, the reality is more nuanced. Professional politicians strike a range of deals in which they retain their agency in mutually beneficial partnerships with criminal organizations. The first major contribution of this book manuscript is to put forth a new conceptual framework characterizing the candidate as a partner who strikes electoral bargains with organized crime, in addition to the more limiting archetypes of pawn, criminal, or partisan. I provide empirical evidence to characterize the nature of such deals, including the terms of trade and broader institutional conditions that make such deals more likely.

Why partner with criminal organizations? The central argument is that hiring criminal groups as political brokers can be a winning electoral strategy for certain candidates. In Section I, I argue that hiring criminal groups as brokers can be an especially effective way for candidates to resolve information asymmetries about where their votes are coming from and how many they will receive. This section explains how candidates mitigate three major risks of entering criminal brokerage deals---physical, legal, and reputational---while on the campaign trail. Section II lays the microfoundations for why criminal groups can be effective brokers. I argue that criminal brokers leverage certain capabilities they derive from control over the population and control over territory to deliver votes to their candidate partners. These brokerage deals help candidates win elections. Section III argues that criminal brokerage deals have downstream implications for the population living under criminal control: first as voters, then later, as citizens. I argue that obtaining votes from criminal brokerage can reduce accountability and perpetuate inequality.

The empirical evidence in this book integrates several original pieces of evidence, collected over twenty months of fieldwork in Rio de Janeiro. This includes over seventy interviews with candidates for local office, community leaders, criminal group members, and voters; an in-person survey administered to more than 350 residents subject to criminal governance; and a telephone survey administered to more than 300 candidates for local office. I also collected information from administrative and NGO records to create an original large-N database linking voting records with criminal governance and politician accountability. To compile these sources, I generated original data, scraping public websites and using text processing tools to prepare the data for analysis. Each of these sources provides important insights into the role of criminal governance in electoral politics. I evaluate my theory using a multi-method strategy that includes multiple natural and survey experiments, observational analyses, and case study comparisons.

This project focuses on local politics in Rio de Janeiro, Brazil for theoretical and empirical reasons. This setting has rich within-case variation in the types of collusive agreements (from long-term partnerships to arms-length spot deals) and types of criminal groups (from warring drug trafficking organizations to extortion rackets). The subnational variation allows me to hold the broader institutional setting constant to isolate the effect of deal-striking on electoral outcomes and eliminate alternative explanations.

My book speaks to broad debates within comparative politics. There is not yet a theory explaining how distributive politics operate amid criminal governance, two critical modalities of politics in the developing world that have largely been examined separately. Existing theories of criminal politics focus on what criminal groups can gain from getting involved in elections and on explaining variation in criminal violence, while leaving their candidate partners in a black box. I depart from these studies of criminal politics by focusing on what politicians can gain from colluding with criminal actors. Finally, this book has implications for broader debates about public security policy. To understand why governments have been largely unsuccessful at constraining criminal groups' power and have allowed violence to escalate, I argue that we must consider candidates' electoral incentives and their relationships with powerful criminal intermediaries.


working papers

Criminal revenue, civic returns: how illicit taxation boosts electoral participation. UNU-WIDER working paper 86/2025. Submitted. [paper]

  • Recipient of the Best Paper Award from the Subnational Politics and Society Section, LASA 2020
  • Previously titled “Criminal capital and voter mobilization”
Abstract How does criminal group taxation affect participation in elections? I argue that criminal groups that tax public service provision use it as a technology of governance, which gives them a comparative advantage in voter mobilization. I predict that higher levels of criminal taxes on services ultimately lead to higher levels of voter participation, and contrast the service provision mechanism with other mechanisms related to coercive taxation and bottom-up reactions to being taxed. I test this argument in Rio de Janeiro, Brazil, where there is significant variation in taxation by drug trafficking gangs and by vigilante groups across the city’s favelas (informal settlements). To measure criminal taxation, I machine-coded text summaries of 1.45 million anonymous calls for help, and construct favela-year-level indices of criminal a) protection taxes and b) taxes on services. My empirical strategy leverages exogenous variation in favela resident presence at the ballot box to estimate the effect of exposure to criminal taxation on voting. I find that criminal taxes on services are strongly associated with higher levels of voter participation, but taxes on protection have the opposite effect. Consistent with my proposed mechanism, I find that taxes on service provision enable criminal groups to be more effective brokers, increasing participation.


Under the radar: how predatory policing distorts the reporting of violence against women. (with Isabella Montini). Submitted. [paper] [pdf] [policy brief]

Abstract A fundamental component of citizenship includes seeking protection from crime and violence. How do citizens make such claims to law enforcement when the police themselves are often predatory? The core argument is that police predation (1) discourages citizens from reporting crimes directly to the police while (2) shifting reporting towards intermediated channels. We argue that trust is the mechanism driving this reporting behavior. To test our theory, we focus on the reporting of violence against women (VAW) in Rio de Janeiro, Brazil. We build an original database of reports to the police and to an intermediary (an independent hotline), then conduct descriptive and causal tests to show that exposure to police violence redirects reporting from the police to the hotline. These findings bring together the claim-making and policing literatures to demonstrate that police violence incentivizes citizens to report VAW to alternative channels and, in their absence, to stay silent.


Criminal correlates of extra-legal taxation: evidence from Brazil UNU-WIDER working paper 89/2026. (with David Becerra-Medina). Under review. [paper]

Abstract Extra-legal taxation is one of many possible tools that organized criminal groups can use to exercise control over territory and populations. Yet the use of extra-legal taxation varies greatly: it is a daily occurrence in some criminally governed communities, and a rarity in others. Where are we more likely to observe extra-legal taxation? This study explores the correlates of extra-legal taxation within the oft-criminally governed favelas of Rio de Janeiro, Brazil. Using novel data from a public security hotline, this paper examines granular, favela-level variation in two types of extra-legal taxation: protection and service taxes. We show that extra-legal taxation is strongly correlated with other forms of organized criminal activity, and is concentrated in the most populous favelas, in the favelas closer to formal transit infrastructure, and in favelas in the lower-middle band of the favela income distribution. In all, this paper contributes to our understanding of the role extra-legal taxation plays in different types of criminal governance regimes and how it interacts with the surrounding state infrastructure.


The electoral returns to vote buying: Evidence from Brazil. (with Almila Basak). [paper]

Abstract Vote buying is widespread, yet we lack systematic evidence about its returns at the candidate-level. We construct a novel database, drawing from more than 30,000 voter tips about candidate malfeasance, to measure the prevalence and efficacy of a range of vote buying modalities across two election cycles in Rio de Janeiro, Brazil. We leverage exogenous variation in voters’ likelihood to file a tip to estimate the effect of vote buying on winning. Our results suggest that buying votes increase a candidate’s probability of victory by 4.49 percentage points, and that the most effective sub-type of vote-buying is the promise of patronage, followed by the distribution of gifts and the use of brokers. While prior work about the persistence of clientelism has focused on citizen demand for benefits, this paper directly connects buying votes to winning through the supply side, underscoring the importance of candidate incentives in explaining why vote buying persists today.


Venue-based sampling: a practical tool to sample hard-to-reach populations in political science [poster]

Abstract Despite an increasing reliance on survey data, finding and sampling understudied populations remains a serious obstacle in political science research. This is particularly true for studies where we lack a high-quality sampling frame, such as conflict or rural settings, or informal areas in the developing world. This paper offers suggestions for how to find and interview these hard-to-reach groups outside of their homes, using a technique called venue-based sampling. I argue that venue-based sampling is underleveraged in political science, not just as a tool to find people, but also as a technique to improve response quality and, under certain conditions, better protect respondent safety. This paper explains the tradeoffs of using venue-based sampling to find people, to elicit truthful answers, and to conduct a safer survey. This paper illustrates these suggestions with evidence from an in-person survey conducted in Rio de Janeiro, Brazil, which used a particular type of venue-based sampling -- worksite sampling -- to interview bus drivers and fare collectors about criminal governance in their neighborhoods.


in progress


Strategic cartography: how map-making enables or obstructs social inclusion (with Ana Paula Pellegrino)

Abstract The poor are often excluded from official maps. Developing countries, particularly their urban areas, have massive inequalities in the quality and comprehensiveness of official maps, affecting the unmapped’s access to downstream social services and resources. How does “putting people on the map” affect inequality? Can low-income maps and registries be used as policy tools to counteract the challenges of living in a low-income settlement?

This project argues that informal settlement registries are a policy tool that can reduce inequality in low-income areas. We substantiate these claims in a two-step nested analysis. First, we conduct a micro-level analysis of 1,074 neighborhoods in Rio de Janeiro, Brazil that were “put on the map” at different points in time. This within-city analysis lets us hold external factors constant and compare development outcomes in different neighborhoods. The rollout of the informal settlement registry began in 1981 and is updated periodically, which also lets us estimate the cumulative effects of “being mapped” over time on inequality. Second, we conduct a meso-level analysis of 152 medium and large cities in Brazil, and the determinants of whether or not they have an informal settlement registry. We use this study to eliminate alternative explanations about the reasons for adoption of an informal settlement registry, and to show how municipal trajectories of poverty and inequality change post-adoption. These two empirical strategies consider variation at the individual, neighborhood, and municipal levels. They show how including informal settlements on the map can change the life trajectory of those living in poverty and be a powerful inequality-reduction tool – if incumbents choose to deploy it.

What electoral behaviors do voters complain about? A new dataset on electoral malfeasance in Brazil [poster]

A call of duty? police force formation in Brazil (with Ana Paula Pellegrino, Joana Monteiro, and Michael Weintraub)