Harnessing the data revolution to fill the poverty data gap
Traditional poverty measurement apporaches often provide insufficient data. Can new data sources fill the gap?
Tackling extreme poverty and monitoring progress towards achieving the Sustainable Development Goals (SDGs) requires access to high quality and timely information. However, traditional approaches to poverty measurement rely on in-person household surveys and census data, and for various reasons these often provide stakeholders with information that is either outdated, unavailable at the right geographical level, or too static. This leaves a poverty data gap.
A DEEP review examines the role that the data revolution can play in filling this gap, with new data sources including remote sensing, the data exhaust, online data, crowd-sourcing, mobile phone surveys, and analytical techniques derived from new technology.
While these new data sources will not replace tried-and-tested approaches to poverty measurement, the two can be used together to provide better evidence on poverty in low- and middle-income countries (LMICs). Data sandwich approaches (approaches that use novel statistical techniques to combine traditional and new sources of data) are a particularly promising avenue.
Our review identifies many studies combining traditional data with data from remote sensors such as satellite imagery, and the data exhaust such as Call Detail Records (CDRs), for high-resolution poverty mapping and targeting. We found fewer studies using other new data sources such as citizen-reported and online data to measure poverty in LMICs.
Our review identifies three future areas for DEEP research:
- Exploring data sandwich approaches further, including investigating whether innovative analytical methods can make high-resolution poverty estimates more precise
- Experimenting with less explored data collection approaches – e.g. crowd-sourcing or mobile phone surveys – to measure poverty in LMICs.
- Investigating whether integrated data systems, comprising both traditional and new data sources, could be a component of national statistical ecosystems for measuring poverty in LMICs. This also implies tackling questions related to ethical concerns, data quality, trust in highly modelled results, and actual demand for new poverty measures.
Read the paper in full: How can new technology support better measurement of extreme poverty?