Evidence Generation
Asia
Evidence Generation
Asia
Latin America is the most violent region in the world, with about 20 homicides per 100,000 inhabitants.1 To combat crime, many governments turn to mano dura (“iron fist”) approaches to domestic law enforcement. These policies frequently include deploying military forces for policing, which typically benefit from strong public backing. However, IPA evidence from Colombia has shown that militarized policing has failed to reduce crime and even exacerbated human rights abuses. Another approach has been to create new, militarized police squads rather than deploying army soldiers.
The challenges facing policymakers across Latin America are especially present in Brazil. Violence in Ceará, a state of almost 10 million people, has risen dramatically over the last ten years. In 2014, the homicide rate in Fortaleza, the capital city, reached 77 per 100,000, the highest among Brazilian state capitals.
Accordingly, to combat the increased violence in Fortaleza, in 2004, authorities created Rondas e Ações Intensivas e Ostensivas (RAIO). RAIO is a militarized, motorcycle-based police unit designed to use their mobility and shows of force to respond quickly and target high-crime areas, especially in dense urban and informal settlements. Intended as a complement to traditional policing, members of this squad were recruited from pre-existing military police forces and selected for their strong human rights records. They received intensive training (280 hours of additional training), better wages (about 30 percent higher than regular officers), and greater supervision. In 2015, the program was rolled out to other municipalities in Ceará in three phases.
In partnership with the Military Police of Ceará, researchers conducted a pilot evaluation to measure whether RAIO improved crime rates, and if so whether it also affected other operational measures like arrests and guns and drugs seized. They leveraged the staggered rollout of RAIO across 184 municipalities over the three phases, in which the municipalities that had not yet or would not receive RAIO acted as comparison groups.
Researchers also administered a survey to 2,000 residents in Fortaleza, asking about the perceived differences between the RAIO and the typical military police. To assess the mechanisms behind the results and to understand what attribute of the RAIO was most influential in shaping resident attitudes, researchers randomized an image of a police officer to have one of each pair of characteristics:
Researchers also sought to understand the impact of the RAIO patrols on police killings, as well as citizen attitudes towards incumbent governments by using official electoral data from the 2018 re-election campaign of a state governor. Finally, they conducted a cost-benefit analysis to understand whether the program's operational cost was justified by the benefits it produced.
Results show that RAIO patrols dramatically reduced crime. Relative to municipalities that had not yet received a RAIO base, homicides fell by 57 percent three years after the rollout, while robberies fell by 84 percent immediately after, with a sustained reduction for four years. Notably, human rights abuses—measured by police killings—did not increase. In the medium term, arrests fell, suggesting that deterrence was likely one of the main mechanisms behind observed crime reductions. They also suggest that intensive training, better working conditions, and higher accountability measures likely are responsible for these changes.
The RAIO patrols were also viewed positively by the public. When asked about effectiveness in reducing the activities of criminal groups, 89 percent of residents felt that the RAIO patrols were at least partially effective, whereas 72 percent said the same about the military police. The image-based conjoint experiment, which provided respondents with images of an officer with randomly-varied attributes, finds that while the motorcycle—signaling rapid response—is particularly important for increasing residents' perceptions of effectiveness, safety, and the likelihood of using force against criminals, the RAIO uniform is far more important in reducing perceptions of abuse and corruption.
The RAIO rollout also boosted electoral support for the incumbent governor, aligning with evidence that successful public safety programs can enhance politicians’ vote shares, especially during election years with intensified program expansion.
Based on these findings from Ceará, researchers conclude that creating new, specialized police units to work parallel to pre-existing police structures may be a viable option where traditional security reform pathways are bureaucratically constrained. These new units may facilitate both better training, enabling officers to respond without violence against citizens in high-stress situations, and better accountability measures, building legitimacy and trust in police forces. Furthermore, a cost-benefit analysis suggests that RAIO’s program costs are substantially lower than the social costs of the crimes it reduces.
Importantly, the researchers emphasize that the RAIO program’s specific training, working conditions, and accountability are likely essential components for positive outcomes when scaling this type of program. They caution against solely attributing the positive outcomes to the militarization of these forces. Though the researchers also recognize that it is not possible to separate the effects of RAIO officer selection of high-performing officers from the superior training they then receive, these results nonetheless offer revelations about core, organizational solutions for police forces in Latin America and beyond.
1. World Bank Data, “Intentional homicides (per 100,000 people) - Latin America & Caribbean,” World Bank Group, Accessed October 29, 2025, https://data.worldbank.org/indicator/VC.IHR.PSRC.P5?locations=ZJ
