Marcelino Guerra
Fortaleza is the fifth largest Brazilian city and is the state capital of Ceará, located in the Northeast region of the country
The city has 120.6 sq mi of territorial area and population around 2.7 million - similar to Houston and Chicago
Census 2010 divides the city into 3,043 census tracts - a.k.a neighborhoods - and 3,020 are populated
In July 2014, The City of Fortaleza began an ongoing urban renewal project called “Areninhas”
The intervention consists of synthetic football turf, a playground and an outdoor gym. The surroundings are improved with further infrastructure development
The project targets vulnerable communities and aims to provide an amenity that promotes physical activity and stimulates a sense of community
In 8 years, 103 football arenas were built in Fortaleza and more than 160 in the rest of the State
There are two types of equipment - Areninhas type 1 and 2. They differ by the size of the turf field and presence of locker rooms and bench
Each equipment have three employees. Two guards (one in the morning, one in the evening) control access to the fields, and one janitor cleans and organizes the place.
According to the City Hall, maintenance costs (salaries, energy and water) range from R$ 120,000 to 125,000/year.
Currently, 102 neighborhoods have an equipment - around 94,000 people covered
City Hall and State Government invest R$ 24.87 millions/year in four social projects: Esporte em 3 tempos, Futpaz, Esporte superação, and Atleta Cidadão
Projects directly impacts around 32,000 children
Amateur football championships are held on weekends, dance/fitness classes occur regularly, and pick-up football games happen at daily basis
There are large disparities in health and safety between communities within a city, and local governments continue to seek meaningful policies to improve residents’ quality of life
Lack of safety in neighborhoods have other detrimental effects, e.g., violence exposure afffects student performance - Sharkey (PNAS, 2010), Monteiro and Rocha (Restat, 2017), Casey et al. (AEA P&P, 2018), and Koppensteiner and Menezes (JOLE, 2021)
Successful neighborhood interventions are likely to generate positive spillovers, and the social benefits of the public policy might be underestimated
This study wants to evaluate an ongoing urban renewal project that blends person and place-based interventions in a large Brazilian city and establish a causal chain between environmental design, violence, and student performance
Murder rates decreased by 2/3 in treated census tracts meaning 62 fewer murders per year
The decrease is larger during afternoon hours and weekdays (when most social projects happen), which supports the “more eyes on the streets”/increase of capable guardians argument - McMillen et al. (JUE, 2019), Sanfelice (JEBO, 2019) - and voluntary incapacitation
Males aged 18-29 with a past criminal record are more impacted. This result suggests that there are fewer gang fights in treated neighborhoods
There are diffusion of benefits to schools up to 500 meters from an equipment. On average, students’ math scores increase by 2.5%.
Social benefits recover all program costs (fixed, maintenance, and social projects) within 2 years
The causal effect of a citywide large-scale 8-years span urban renewal policy on violence and student achievement
Exploiting the time of murders, I am able to discuss some mechanisms through which the intervention i) is curbing violence ii) is improving student performance
Assessment of the policy’s indirect impacts. Due to the research design/intervention’s characteristics, few studies can correctly identify temporal and spatial displacement/diffusion of benefits
The intervention takes place in a large and growing city in a Developing country
Heterogeneity analysis shows that the intervention has a larger effect on the most vulnerable population: young males with a past criminal record
The study supports a combination of place and person-based interventions to prevent public goods from turning into public bads - Albouy et al. (JPUB, 2020)
Fortaleza-CE commonly ranks as one of the most violent cities in the world, with homicide rates close to Cali-COL, St Louis-USA, New Orleans-USA, and Baltimore-USA
Between 2004 and 2015, homicide rates more than double. Many factors contributed to the violence escalation: from a police strike at the end of 2011 to the rise of crime syndicates
In 2019 the city managed to cut murder rates by more than 50%, compared to the previous year. However, in February 2020, the Military Police went on a general strike again, and the State went back to the top of the murder rates ranking
Detailed information about 12,081 murders within Fortaleza’s boundaries between January 1st, 2012 and November 30th, 2019
The 8,291 homicides that happened on the streets are considered in the analysis
5,560 murders occurred between 8:00 am and 10:00 pm
Murders are aggregated to months and at the census tract level. The outcome is homicide rates per 1,000 people
Note: complete addresses were converted to coordinates (lat and lon) using Google’s API
Since 1992, the State Government of Ceara has been consistently evaluating the performance of students in the public system. Every year, middle and high school students take the SPAECE exam by late November
The SPAECE contains standardized Math and Portuguese test scores - it is an average at the school level
School Census data have yearly information about schools’ characteristics (internet coverage, presence of labs/library, number of employees, etc.)
I conjecture tracts treated until July 2019 and after January 2020 are similar in observables and unobservables, and this control group’s choice produces an apples-to-apples comparison
Potential control areas within 650 meters of a treated neighborhood are excluded (2), as well as fields built between September and December 2019 (11)
Different comparison groups are also built by matching similar never treated neighborhoods using neighborhoods and boroughs’ demographics available in Census 2010
In total, there are 31 treated and 55 control neighborhoods
Total population in treated areas is 28,365 and in control tracts is 51,017
This sample covers 3.2% of the city’s population and around 6% of land area
I draw rings with [300m, 600m] radius from 86 quasi-experimental neighborhoods to identify treated and control schools
Any overlap between treated and control is kept as treated
If a school is covered by more than one treated area, the school will be attached to the arena with the earliest opening date
If a school is covered by more than one control area, duplicates are dropped based on the minimum distance to the equipment
ETWFE point estimate suggests that the urban renewal policy caused a decrease in murder rates decreased around 2/3 in treated neighborhoods
S & A (2021) confidence intervals rules out reductions smaller than 26%
The figure shows ETWFE and S&A (2021) dynamic treatment effects. Months are binned into years, and endpoints are binned at -4 years (or less) and 2 years (or more). Coefficients are normalized to event time -1.
Day-of-week columns show a larger effect during weekdays, and time-of-day analysis points to a bigger effect during the afternoon (from 12 pm to 5:59 pm) - when most social projects happen
Pupils that study during the afternoon and commute from home to school around 1:00 pm and from school to home between 5:00-5:30 pm would benefit the most
The effect is driven by males aged 18-29 with past criminal record, which suggests fewer gang fights in treated areas
Although there is anecdotal evidence of young males leaving the drug trafficking in these areas, incapacitation effect is hard to identify
The first three columns display the causal effects of the urban policy on Portuguese scores of students in \(5^{th}\) grade. Results point to an increase of 0.56%, but are not statistically significant
Using the preferred ring size, Math scores increase by 2.5% on average
As the radius increases, point estimates get smaller - i.e., schools very close to sports arenas (below 300 meters threshold) benefit the most
At such small range, inference suffers from few treated clusters, and confidence intervals are made using wild bootstrap
Back-of-envelop calculation shows that the 31 treated areas are having 18.86 fewer homicides every year. Extrapolating the effects to all 102 areas covered by 103 football fields, there are 62.68 fewer murders/year due to this policy. I follow Pereira, Almeida and Oliveira (2020), who attach R$ 1.119 million for a blue-collar Brazilian men
Amenity value is based on private turf field rentals at an hourly rate of R$ 100, considering 56 hours/week available for practice
Building costs vary from R$ .24 to R$ 1.7 million depending on the type of the equipment, and maintenance costs are composed by salaries, water and energy. There are four social projects connected to areninhas with an estimated cost of R$ 24.87 million/year
Crime prevention through environmental design might be a cost-effective strategy to reduce violent crimes and an alternative to policing
Successful interventions may cause diffusion of benefits to other areas such as school achievement
Place-based policies that involve the community or significantly change the neighborhood’s routine might have a greater impact on residents’ quality of life and prevent public goods from turning into public bads