Statistics do not yet show a "safer Rio" post-UPP installation

Lately, it seems as though there has been a surge in violent crime and armed robberies in neighborhoods with UPPs. As the media in Rio is dominated by pro-government management, one can only assume that the crimes which newspapers - such as O Globo - choose to report represent a mere fraction of the crimes which actually transpire.

Regardless, Beltrame and the public security forces continue to insist that the UPPs have made major strides in reducing crime in Rio de Janeiro. In particular, the Secretary of Public Security likes to pepper their proof-positive discourse by citing reductions in car theft, homicide, petty crime, and armed assault, often without producing any corroborative statistics. Emphasis is usually placed on the benefits conferred by the UPPs on the those living on the asfalto - that is, the wealthier "legal" neighborhoods surrounding the comunidades.

The general public - especially those comprising Rio's middle class and above - tends to accept that there is a causal relationship between UPPs and crime reduction. Even within the comunidades, the UPPs enjoy some support; residents praising the units are quoted on almost a daily basis by media outlets.

Morro da Coroa borders the touristy Santa Teresa neighborhood.
Coroa's UPP was installed earlier this year. Photo credit: UPPRJ

Others are less optimistic, and postulate that the UPPs have simply "displaced" crime to other municipalities - and even other states. See this article for an overview.

However, by any statistical measure, it is far too soon to demonstrate a correlation between UPP installations and crime reduction/displacement; there simply is not enough available data (the first UPPs were installed less than three years ago). Furthermore, outside of the laboratory, in which one can control for undesired independent variables, it is nearly impossible to prove causality. Any statistician who wishes to show a causal relationship between increased UPP presence and crime reduction would be hard-pressed to substantiate his hypothesis.

In Rio de Janeiro, where accurate, consistent, and impartial data sources can be hard to come by, the challenge of demonstrating causality becomes even more daunting, and claims of UPP success appear all the more dubious. The two main sources of data on crime are the Brazilian Institute of Geography and Statistics (IBGE), and the Public Security Institute (ISP). Both institutes are run by the state government, are notoriously difficult to to navigate, and generally do not publicize public security data prior to the year 2000.

However, even given these obstacles, it is still possible to produce an initial picture of the relationship between UPPs and crime. Using the IPS database, I performed a simple bivariate analysis using data from the regional, city, zonal, and neighborhood levels, using the number of UPPs as the independent variable and categories of crime as the dependent variables. As always, there are many limitations to the data; besides the obstacles described above there were also issues with changes in data collection methodology, categories, and presentation.

At the regional level, I took data from the City of Rio de Janeiro, the Baixada Fluminense, Grande Niteroi, and the Interior. At the zonal level, I examined the Zona Sul, Centro, Zona Norte, and Zona Oeste. At the neighborhood level, I used the examples of Copacabana/Leme, Realengo/Bangu/Anchieta, and Tijuca/Vila Isabel/Maracana (in the South, West, and North zones, respectively). The basis for the choices of neighborhoods was that the sample would feature a wealthy neighborhood with an older UPP* (Copacabana/Leme), a lower-class neighborhood with an older UPP (Realengo/Bangu/Anchieta), and a middle-class neighborhood with a more recent UPP* (Tijuca/Vila Isabel/Maracana).

Population growth rates were taken into account, and all analysis is based on the most recent census data (2010). Crime data was taken from the years 2005-2011, as available. Adjustments in IPS's data collection methodology preclude the examination of several categories of crime prior to 2008, which will be addressed later.

If someone held me to gunpoint right now (add one incidence on to the "assault" category, please), I would have to say that there is very little evidence that suggests that the city of Rio de Janeiro has become safer.

Stay tuned for the cold, hard data tomorrow. As a teaser, I'll leave you with this scatterplot of homicides in Rio de Janeiro. Note that homicides began to plummet well before the installation of the city's first UPP in late 2008, and actually increased in 2009 before leveling off to pre-UPP levels in 2010.

*Copacabana/Leme and Tijuca/Vila Isabel/Marcana neighborhoods now have 2 and 5 UPPs, respectively