Body cameras on police uniforms: the case of Sao Paulo
Technological advances in capturing and storing images have allowed police forces to adopt innovative approaches to improve their performance and increase citizens’ confidence in their work. In particular, body cameras on police uniforms have been used in countries such as the United States and England at least since 2005 and this phenomenon recently gained impetus following episodes of excessive police violence in various countries around the world. In Brazil, the Sao Paulo State Military Police introduced body cameras to their units in June 2021, through the “Olho Vivo” (“Keen Eye”) program, part of broader efforts to reduce killings by law enforcement officers.[1]
What has been the impact of introducing body cameras in Sao Paulo?
Since the Keen Eye program began, indicators for the use of police force have shown a significant drop. In the units that received the cameras, the total number of deaths resulting from police intervention fell 80% compared to the previous 12 months.[2] Although significant, this decline does not allow us to ascertain the full impact on the use of force.
To assess the benefits of any public policy accurately, it is important to have a good estimate of the counterfactual, meaning the hypothetical scenario in which cameras were not used but all other determinants of police killings evolved as they did in reality. In many cases, it is not possible to estimate this alternative scenario due to the absence of an appropriate control group. In this case, however, the phased rollout of body cameras across different Sao Paulo police units means that the units that did not receive the cameras are a good control group for the counterfactual trajectory of the units that received them.
In our article “Avaliação do impacto do uso de câmeras corporais pela Polícia Militar do Estado de São Paulo” (“Evaluation of the impact of the Sao Paulo State Military Police’s use of body cameras”), we used econometric techniques to rigorously evaluate the effect of the introduction of cameras on the use of police force. The results indicate that, on average, the cameras reduced the number of police killings by 0.22 every two months, compared to the units that did not receive the cameras. This represents a reduction of 57% in relation to the average level of this indicator prior to the Keen Eye Program. In other words, the decline is smaller than the 80% figure obtained without using robust statistical techniques, but it is still a significant reduction. In addition to deaths, injuries resulting from police intervention also fell significantly, by 61%.
Why do these cameras reduce the use of force?
There are at least three ways in which body cameras may reduce the use of force. First, it is possible that body cameras make police officers more wary of approaching suspects in order to avoid mistakes that could lead to punishment, reducing their level of proactivity and effort. However, the trend in the number of arrests did not differ between the units that received the cameras and the other ones. Moreover, the number of crimes such as robberies, thefts and homicides detected did not change because of the introduction of the cameras. So, there is no evidence that this public policy has incurred a cost in terms of increased crime in Sao Paulo.
The second possible reason for the decline in police violence is changed behavior among citizens and suspects when interacting with police officers. It is possible that, upon learning that they are being filmed, people may become less hostile toward the police, reducing the need for forceful action by law enforcement officers. However, looking at data on people who resist arrest, we found limited evidence for this hypothesis. The introduction of body cameras has not caused a significant reduction in this type of occurrence.
Finally, the existence of supervisory mechanisms may generate a change in police officers’ behavior, aligning their conduct with police protocols. This is a particularly hard hypothesis to test, as it is necessary to use microdata about videos watched by the police oversight structure. However, this hypothesis is supported by an increase in incident reports observed in internal police records. The cameras induced a 12% increase in total records. The results were particularly strong regarding incidents that have typically been underreported, such as domestic violence, whose records increased by 102%. Thus, this increase in the use of institutional instruments for reporting incidents suggests that the use of cameras affects police officers’ behavior.
In fact, there are cases in which the use of cameras without institutional support did not generate any changes in behavior, as police officers themselves often did not turn on the devices when necessary.[3] In the case of Sao Paulo, on the contrary, body cameras were adopted after at least eight years of institutional work, which began with technology appraisal, academic studies on the subject and liaison with other police forces. The cameras were finally implemented in 2021, after pilot projects were carried out. Thanks to this thorough process, the police adopted protocols for using cameras in accordance with international best practices, such as the obligation to keep cameras on throughout shifts and active supervision strategies.
The results therefore suggest that body cameras have an effect on reducing police violence, mainly by inducing officers to comply with existing institutional protocols and instruments. However, cameras alone are not capable of solving problems in the area of public security, including excessive force and misconduct by police officers. Their effect on police behavior and compliance with internal protocols is enhanced by the introduction of oversight systems and the existence of accountability mechanisms.
[1] There were also some smaller-scale initiatives in Santa Catarina and Rondônia in 2019.
[2] Source: https://noticias.uol.com.br/cotidiano/ultimas-noticias/2022/07/05/cameras-no-uniforme-da-pm-letalidade-policial-intervencao-lesao-corporal.htm (accessed on February 7, 2023)
[3] Magaloni, B., Melo, V., Robles, G., & Empinotti, G. (2019). How body-worn cameras affect the use of gunshots, stop-and searches and other forms of police behavior: A Randomized Control Trial in Rio de Janeiro.