Cross-analysis of public-sector compliance with Peru's 5% disability employment quota (Law 29973) against the territorial distribution of working-age people with disabilities. Produces three composite indices — coverage, viability, and intervention priority — mapped at the institutional level across all Peruvian districts.
Data sources: DFS (Employment Quota Registry, Jan–Dec 2023) and DPRIPD (PCD Registry by age group and severity, Sep 2024) — both published by GeoPeru / CONADIS.
Peru's Law 29973 requires public institutions with 20 or more employees to reserve at least 5% of their workforce for persons with disabilities (PCD). Despite this mandate, enforcement data is fragmented across four registries and has never been systematically cross-referenced with the actual supply of employable PCD at the district level.
This project integrates both sides of the equation — institutional demand and territorial supply — to produce an actionable, georeferenced picture of where the gap is largest and where it is most feasible to close.
- Quantify the compliance gap between the legal quota and actual PCD employment in each public institution.
- Estimate the pool of working-age PCD (mild/moderate disability) available within each district.
- Derive three composite indices: coverage, viability, and intervention priority.
- Produce publication-quality choropleth maps at the national level with a Lima Metropolitan inset.
- Publish results as an interactive dashboard via GitHub Pages.
| Property | Detail |
|---|---|
| Source | GeoPeru / CONADIS open data portal |
| DFS period | January – December 2023 |
| DPRIPD period | September 2024 |
| Geographic level | District (distrito), province, and department |
| Institutions covered | Regional governments, provincial municipalities, district municipalities, and other public entities |
| Key variables | TOT_TRAB, TOT_PCD, FALT_PCD, PCD_PEA, severity levels L0–L3, UBIGEO codes, coordinates |
| Format | .dbf inside .zip archives (downloaded directly from remote URLs) |
- Data acquisition — Remote download of 9 ZIP archives from GeoPeru;
extraction of
.dbffiles in memory with automatic latin-1 → UTF-8 encoding correction. - Cleaning & validation — Column selection, type casting to nullable
Int64, clipping of out-of-range ratios. - Transformation — PEA proxy computed from age cohorts 18–59; severity
levels merged by UBIGEO; working-age employable PCD (
PCD_PEA_L1L2) weighted by mild/moderate share. - Index construction
IDX_COBERTURA= min(actual PCD / quota, 1)IDX_VIABILIDAD= 1 − min(deficit / district supply, 1)IDX_PRIORIDAD= (1 − coverage) × viability
- Spatial analysis — Four DFS tables unified; institutions projected to
Web Mercator (EPSG:3857); point-layer built with
GeoPandas; country mask derived from Natural Earth shapefiles. - Visualisation — Three choropleth maps (one per index) with
proportional markers scaled by
FALT_PCDand a Lima Metropolitan inset; exported as transparent PNG at 150 dpi. - Publication — Dashboard deployed on GitHub Pages from
index.html.
geoinclusion/
├── data/ # Raw .dbf files (not tracked — see .gitignore)
├── docs/ # Reports
│ └── Propuesta_de_Solucion.pdf
├── img/ # Generated maps and charts (tracked)
│ ├── mapa_IDX_COBERTURA.png
│ ├── mapa_IDX_VIABILIDAD.png
│ └── mapa_IDX_PRIORIDAD.png
├── notebooks/ # Main analysis notebook
│ └── notebooks/analisis_cuota_empleo_pcd_geoperu.ipynb
├── .gitignore
├── index.html # Interactive dashboard (GitHub Pages)
├── README.md
└── requirements.txt
Measures how close each institution is to meeting the 5% quota. A value of
1.0 means the quota is fully met; values below 0.5 indicate that fewer
than half the required PCD positions are filled.
Estimates whether the district where the institution is located has enough working-age PCD (mild or moderate disability) to absorb its unfilled positions. High viability means the labour supply exists locally.
Composite score that ranks institutions by intervention urgency: highest priority is assigned to those with the largest quota gap and the highest local labour supply — i.e., where non-compliance cannot be attributed to scarcity of candidates.
| Coverage of the 5% quota | District-level PCD labour supply | Intervention priority |
|---|---|---|
 |
 |
 |
Marker size is proportional to the number of PCD positions still unfilled
(FALT_PCD). Inset shows Lima Metropolitan area at higher zoom.
An interactive version of the results is published via GitHub Pages:
View the Geoinclusion Dashboard
- A significant share of evaluated public institutions fall below 50% of the required quota, with the largest absolute deficits concentrated in large provincial municipalities and regional governments.
- Several districts combine high non-compliance with a substantial pool of working-age PCD at mild/moderate severity — indicating that the gap is structurally addressable.
- Lima Metropolitan concentrates the highest number of high-priority institutions, though non-compliance is spread across all regions.
- Temporal mismatch: the DFS employment data covers 2023 while the DPRIPD disability registry corresponds to September 2024; the cross-sectional comparison is approximate.
- PEA proxy: working-age PCD is estimated from age cohorts rather than observed labour-force participation rates.
- Geocoding coverage: institutions lacking valid coordinates are excluded from maps but are included in tabular analysis.
- Severity weighting: only mild (L1) and moderate (L2) cases are counted as "employable", which may underestimate true supply.
- Registry completeness: self-reported figures in the DFS may underreport actual PCD employment.
- Incorporate the 2024 DFS update when published by CONADIS.
- Add formal labour-force participation rates by disability type (INEI ENEDIS survey) to refine the viability index.
- Build a district-level ranking table exportable to CSV/Excel for operational use by enforcement agencies.
- Migrate visualisations to an interactive web map (Folium or Plotly) for finer exploration.
- Automate the pipeline with a scheduled script to refresh data on each GeoPeru release.
Analytical proposals and policy recommendations derived from this work are presented in a separate document:
Read the full Proposals and Policy Recommendations Report (PDF)
# 1. Clone the repository
git clone https://github.com/<your-username>/disability-employment-quota-peru.git
cd disability-employment-quota-peru
# 2. Create and activate a virtual environment (recommended)
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
# 3. Install dependencies
pip install -r requirements.txt
# 4. Launch Jupyter and run the notebook
jupyter notebook notebooks/geoton_2026_clean.ipynbNote: the notebook downloads all data directly from GeoPeru's public URLs — no manual data download required. A stable internet connection and ~200 MB of free memory are sufficient.
| Library | Role |
|---|---|
pandas 2.x |
Data wrangling and tabular transformations |
numpy 1.x |
Vectorised arithmetic and ceiling/clip operations |
geopandas 0.14+ |
Geospatial data structures and CRS reprojection |
shapely 2.x |
Polygon operations (union, intersection, difference) |
matplotlib 3.x |
Static choropleth map rendering |
contextily 1.x |
Web basemap tiles (CartoDB Positron / Voyager) |
requests 2.x |
Remote ZIP download |
dbfread 2.x |
Parsing of .dbf binary files |
jupyter / nbformat |
Interactive notebook environment |
Coco Benel Computational Physicist · Data Analyst · Data Scientist
Junior Alegre Computational Physicist · Data Analyst · Data Scientist
Data published by CONADIS / GeoPeru under open government data licence. This repository is an independent analytical work and does not represent the official position of any government agency.