Mapping poverty using mobile phone and satellite data.

Area of Science:

• Socioeconomic Determinants of Health
• Geospatial Analysis
• Development Economics

Background:

• Poverty significantly impacts global health outcomes, societal stability, and human potential.
• Traditional poverty measurement relies on census data, often unavailable or outdated in low- and middle-income countries (LMICs).
• There is a critical need for alternative methods to estimate poverty between censuses.

Purpose of the Study:

• To demonstrate the utility of public and private data sources for understanding poverty's spatial distribution in LMICs.
• To evaluate the effectiveness of mobile operator and geospatial data in modeling poverty measures.
• To assess the potential for frequent, high-resolution poverty estimation.

Main Methods:

• Modeling three traditional poverty measures using aggregate mobile operator data.
• Incorporating widely available geospatial data into poverty models.
• Stratifying models for urban and rural areas to assess data advantages.

Main Results:

• Models combining mobile and geospatial data achieved the highest predictive power (r² = 0.78) and lowest error.
• Models using only mobile data offered comparable results, indicating potential for frequent, granular poverty measurement.
• Mobile data proved particularly advantageous in urban settings, with different data types showing value in distinct contexts.

Conclusions:

• Public and private data, especially mobile data, can provide novel insights into poverty distribution in LMICs.
• These methods offer a viable alternative for continuous, high-resolution poverty monitoring where traditional data collection is limited.
• The findings support the development of more responsive and accurate poverty assessment strategies.