Speaker
Description
This paper introduces a new method for estimating and mapping the size and spatial distribution of the middle class, leveraging geospatial data and machine learning techniques, and provides the first granular overview of the middle class in Africa. Existing middle-class definitions, based on income or expenditure thresholds and typically estimated using survey data, have never been available with geographical precision in developing contexts due to limited data. Using geolocated survey data as ground truth, we integrate different geospatial information—including infrastructure and nightlight data—mapped onto a hexagonal 96 km² grid (≈6km circumradius). Machine-learning algorithms enable extrapolation to regions without direct survey coverage.
| Keyword | Poverty and Inequality |
|---|
