Case study · Policy brief · 2025
Blackpolis: mapping structural racism in Cali with census data
A policy brief addressed to Cali's equity officials. Using the 2018 census for 328 neighbourhoods, I measured how the share of Afro-Colombian residents relates to unemployment, schooling and illiteracy, then grouped neighbourhoods so that a programme could be targeted where the gaps are largest.
- Data
- DANE 2018 census · CaliEnDatos
- Unit
- 328 neighbourhoods
- Methods
- OLS · clustering · QGIS
What the census shows
For each neighbourhood I took the share of residents who identify as Black, Afro-Colombian, Raizal or Palenquero (NARP) and regressed four outcomes on it. Every slope is significant at p < 0.001:
- Unemployment: +0.30 per point of Afro share (R² 0.34).
- School attendance, ages 5 to 21: −0.55 (R² 0.15).
- Higher-education attendance: −1.09 (R² 0.33).
- Illiteracy: +0.11 (R² 0.35).
Three kinds of neighbourhood
Clustering split the city into three groups. In the most disadvantaged one, 21% of residents are Afro-Colombian, unemployment is 24% and 13% attend higher education. In the most advantaged, 6% are Afro-Colombian and 54.7% attend higher education. The maps, made in QGIS, show the clusters as contiguous areas of the city.
Recommendation
I scored three options against the same criteria: the status quo (18 points), an education programme (21) and a labour programme (23). The brief recommends labour-market entry and personal-finance programmes, targeted at the neighbourhoods in the disadvantaged cluster.
The analysis is in Python (pandas, scikit-learn, plotly): blackpolis_replication on GitHub.