Better poverty mapping: New machine-learning approach targets aid more effectively

ITHACA, N.Y. – Leveraging national surveys, big data, and machine learning, Cornell University researchers have developed a new approach to mapping poverty that could help policymakers and NGOs better identify the neediest populations in poor countries and allocate resources more effectively.

To eliminate extreme poverty, defined as surviving on less than $2.15 per person per day, governments and development and humanitarian agencies need to know how many people live under that threshold, and where. Yet that information often is lacking in the countries that need help most, the researchers said.

Household surveys on income or consumption – considered the gold standard for defining poverty lines – may be unavailable or outdated because they are expensive and difficult to administer frequently. Meanwhile, data from satellites and other Earth observation systems monitoring infrastructure, natural conditions, and human behavior has been successfully used to generate asset-based poverty indexes disconnected from the monetary measure most relevant to policymakers.

The Cornell team’s new structural poverty estimates seek to address that gap by translating abundant Earth observation data into more actionable terms for policymakers.

Focused on four southern and eastern African nations, the pilot project mapped poverty about as accurately as existing asset index methods, but for more useful measures – including the share of people living below the global poverty line. The structural poverty approach outperformed previous monetary poverty methods and is forward-looking, making it especially useful for informing programming.

“Rapid advances in data science haven’t gained widespread acceptance because they haven’t produced usable estimates,” said Chris Barrett, professor of applied economics and management. “We’ve made computational advances more practical by linking them to monetary poverty lines.”

Barrett is the senior author of “Microlevel Structural Poverty Estimates for Southern and Eastern Africa,” published in Proceedings of the National Academy of Sciences as part of a series of inaugural articles by academy members elected in 2022, including Barrett.

The research focused on Ethiopia, Malawi, Tanzania and Uganda – agricultural nations with high poverty rates where many development agencies are working, but with only a rough idea of where the poorest people live, the researchers said.

“These are places where we think the structural poverty model is quite relevant,” said first author Elizabeth Tennant, a research associate in economics. “They’re also places where we had good data on consumption and assets, so we were able to look at both and model their connections.”

The team trained machine-learning models using 13 national household surveys conducted in the four countries between 2008 and 2020, linking them to Earth observation data on assets like housing quality, land, livestock, vehicles, and access to technology including cell phones. In short, the researchers said, the older survey data trained models to generate localized “nowcasts” of current conditions from recent satellite observations.

“We’re showing that you can get all the computational precision of advances the data science community has made, while having the policy and programming usefulness of these monetary measures – and in a forward-, not backward-looking way,” Barrett said. “You want to know who’s expected to be poor right now – not when a big survey was conducted years ago – and that’s what our structural poverty models help predict.”

The research was funded by the Cornell Atkinson Center for Sustainability and received computing support from the Cornell Center for Social Sciences.

For additional information, read this Cornell Chronicle story.

-30-