10 January 2020

Killing in the Slums: The Problems of Social Order, Criminal Governance, and Police Violence in Rio de Janeiro

By Beatriz Magaloni, Edgar Franco, Vanessa Melo


Conflicts related to organized crime and gang turf wars have emerged as the deadliest form of violence in the world. This paper characterizes the variety of forms of local governance that OCGs establish in the territories they control. It also explores the mechanisms and processes that allow the police to take back territorial control and generate legitimate state order, as well as alternative conditions where police interventions fail, leading violence to escalate. Our paper contributes to various bodies of literature including drug trafficking violence, criminal governance, urban crime, policing, and state-building.



This paper generates knowledge about one of the most important security interventions in

Latin America: Rio de Janeiro’s "Pacifying Police Units" (UPPs). Inspired by strategies of community-oriented policing, the UPPs sought to abandon prior militaristic approaches to policing favelas (slums) under the control of drug gangs and paramilitary groups. This paper takes advantage of the staggered implementation of the UPP when over 10,000 police officers were deployed to approximately 160 favelas to understand the challenges states confront in their efforts to regain territorial control.



A world without police is not necessarily one of anarchy. OCGs often establish local forms of governance where they contain violence and sanction crimes such as rape, assault, or robbery. While OCGs maintain order while violently confronting the state under Insurgent regimes, they collude with law enforcement under Symbiotic ones. Police interventions in these settings are likely to disrupt the existing criminal governance that keeps violence under control. Moreover, in Insurgent regimes, community-oriented policing approaches can produce paradoxical results—for instance, by increasing the number of police officers killed—and turn these territories into active war zones. 


By contrast, state crackdowns are likely to improve local security where OCGs are unable to restrain their violence or where they predate on residents, such as in Predatory, Bandit, and Split regimes. Under Predatory regimes, OCGs collude with the state to predate on residents—e.g., they extract heavy taxes from residents through extortion. Meanwhile, under Bandit and Split regimes, they are either unable to restrain their armed men or contest territory with other OCGs, thus leading them to behave more like ‘roving bandits.” In these territories, the police often emerges as a critical player for restraining criminal predation and violence. Nonetheless, states will seldom intervene in Predatory regimes where criminals, police, and state officials jointly profit from the OCG’s illicit extraction of rents. 



This paper proposes a novel way of conducting research on criminality that combines extensive ethnographic research, automated text analysis of crime reports, a large-N survey, and quasi-experimental statistical modeling using crime indicators. To explore variation in criminal regimes, we first use ethnographic research in six territories selected according to our typology. To generalize beyond our case studies and offer support for our theory, the paper uses automated text analysis of thousands of anonymous tips collected by an independent Brazilian NGO, Disque Denuncia


To demonstrate the heterogeneous effects of the UPPs, the paper uses manually geo-coded crime data to contrast treated and non-treated favelas through a difference-in-difference statistical methodology. Territorial control is measured in terms of three quantifiable outcomes: 

  • reductions in armed confrontations between OCGs and the state, measured by the number of killings caused by police officers;
  • improvements in public security, captured by homicides rates and other common crime indicators; and
  • community acceptance of the police, measured by a large (N = 5,300) survey collected in favelas.



  • As predicted by our theory, the UPP reduced homicides and fatal police shootings in Split and Bandit regimes.
  • In Insurgent and Symbiotic criminal regimes, the UPP increased lethal violence and common crime.
  • As predicted by our theory, the community-oriented police strategy also increased fatal shootings and the number of police officers killed in Insurgent criminal regimes.
  • The state refrain from targeting Predatory criminal regimes controlled by Rio’s paramilitary groups. Only one UPP was deployed to a militia-controlled territory where it reduced extortion, as predicted by our theory.
  • The results of our large-N survey demonstrate that where residents felt safer under the rule of OCGs, residents prefer that the UPP leave their favela.
  • In contrast, the community is more accepting of the police in areas where local criminal groups were unable to restrain their violence.
  • Our survey data further demonstrates that abusive policing practices—assault, killings of friends and relatives, and home invasions—dramatically reduces community acceptance of the UPP.
  • Overall, our results demonstrate that the state’s capacity to regain territorial control is jointly influenced by the pre-existing criminal regime and police behavior.



  • Shifting territorial control ultimately requires community acceptance of the police rather than the OCG as the legitimate embodiment of physical force.
  • When criminal rule effectively provides local security, the state will have a hard time gaining territorial control because residents often feel safer under the rule of drug lords than with the police presence.
  • By contrast, where OCGs are unable to restrain their men from fighting among themselves and victimizing residents, it is significantly easier for the state to regain territorial control and create a legitimate state order.
  • Our paper also suggests that when the state fails to sanction police officers’ abusive behaviors and public justice systems fail the poor, the state will likely fail to regain territorial control.



Magaloni, Beatriz, Edgar Franco, and Vanessa Melo (2020) “Killing in the Slums: Social Order, Criminal Governance and Police Violence in Rio de Janeiro.”American Political Science Review 114 (2): 552-572.