Working Paper
Mapping Africa’s Infrastructure Potential with Geospatial Big Data and Causal ML
Using rich geospatial data and causal machine learning, this paper maps potential economic benefits of incremental investments in all major types of public and economic infrastructure across Africa. These ’infrastructure potential maps’ cover all populated areas in Africa, at a spatial resolution of 9.7km (96km2 hexagons). They show that the local benefits of additional infrastructure are highly variable and context-specific. Investments in education, paved roads, power, communications, industrial facilities, health, automotive facilities, transport, public services, and hotels yield high average but very heterogeneous returns. Power investments, for example, significantly increase household wealth only in urban spaces, whereas education investments are more effective in rural areas. Market access and agglomeration effects are important forces governing these returns. Descriptive analysis further reveals that infrastruc- ture in Africa is concentrated in urban areas and often inefficiently allocated. African cities exhibit marked heterogeneity in both public infrastructure and economic activities.
Key Words
- Africa
- Infrastructure
- Investment Potential and Impact
- Geospatial
- OSM
- Causal ML
- Explainable AI