Building each Pinkerton Crime Index necessitates in-depth learning on communities, government and law enforcement, and the conditions of crime and crime data nationwide.
Brazil spans more than three million square miles and is home to over 211 million people. Its population is concentrated in major urban centers like São Paulo, Rio de Janeiro, and Brasília. Within these cities, approximately 16 million people live in favelas — densely populated informal settlements often characterized by high poverty and poor infrastructure services. Meanwhile, about 12% of Brazilians, or roughly 25 million people, live in rural areas.
Crime patterns in this South American country vary significantly across these settings — urban centers, rural communities, and especially favelas. Each environment presents distinct social dynamics, economic pressures, and law enforcement challenges in Brazil. Favelas often require specialized modeling approaches due to their unique structural and institutional characteristics.
Routine Activity Theory Crime Modeling
Guided by Routine Activity Theory, our analysis focuses on how offenders, victims, and capable guardians interact across these contexts. We pay special attention to the spatial patterns and daily routines that influence criminal behavior, tailoring our crime modeling, analysis, and strategies to reflect the distinct realities of each setting. Understanding Brazilian crime and crime statistics, therefore, demands a nuanced view of place — where routine activities operate differently in rural towns, urban neighborhoods, and favelas, and additionally how these dynamics operate across the boundaries between these places.
Multi-source Crime Data Integration
There are several arms of Brazilian law enforcement at the federal, state, and municipal levels. Each of these agencies produces official crime data at different geographic resolutions (precinct, municipal, state, and federal) and at varying temporal intervals (monthly, quarterly, and annually). This fragmented reporting system creates significant analytical challenges. There are gaps in coverage — missing years, missing municipalities — and discrepancies across datasets.
Crime types may disappear from local reporting, or local totals may not aggregate cleanly to state or national levels. These inconsistencies make it extremely difficult to build a coherent picture of crime patterns from the raw data alone. By blending these sources, reconciling inconsistencies, and triangulating data across regions where patterns are consistent, we construct stable time series that capture how crime evolves over time.
In addition to these official sources, we used multiple victimization surveys to investigate shares of underreported and missing crime. Similarly, we investigated alternative data sources from the Ministry of Health to supplement official policing data concerning homicides. To improve our surveillance of motor vehicle theft, we leveraged insurance company and fleet data. Data from Federal highways, the Brazilian Forum of Public Security, non-government organization datasets on gun violence, and more helped inform community characterizations and our understanding of crime risk at the neighborhood scale.
Understanding Crime Seasonality, Population Density, Tourism, and Floating Populations Crime Risk
We used tourism data to extend our understanding of crime seasonality and population in-flows beyond official residential data. Because of the population density of some large subdistricts, home to more than 20,000 people, we built meaningful neighborhoods out of census sectors. To further inform neighborhood characteristics, we traced movements of populations across communities to assess where people live, work, and recreate. Understanding these floating populations informed how we assessed localized crime risk and the balance of suitable targets, motivated offenders, and guardianship on the neighborhood level for more than 30,000 neighborhoods and 99% of Brazil’s population.
Beyond Official Baseline Data: Risk Intelligence and Advisory
The resulting Brazil Pinkerton Crime Index offers a unique value proposition: an accurate, nationwide tool, facilitating comparisons of violent and property crime risk across a diverse range of communities. Many crime and risk products do little more than present official baseline data. The Pinkerton Crime Index drills down further, assessing many data sources to reconcile uncertainty, elegantly crafting statistical machinery checked against real-life risk expertise, to provide salient crime risk insights useful to decision makers and stakeholders.
