Combining Satellite Images and Artificial Intelligence to Measure Poverty in 1982-2020

Ouagadougou, Burkina Faso 1986-2006: The two images illustrate how the urban material change correlated with poverty trends that our algorithms will quantify.

Status:

Ongoing

 

Date:

2021-Present

 

Overview:

In the Spring of 2021, Ellen Lust, Adel Daoud (Department of Sociology, GU – Principal PI), Dr. Fredrik Johansson (Chalmers), Dr. Maria Brandén (Linköping University), Professor Xiao-Li Meng (Harvard University), Professor Peter Hedström (KTH), and Professor Devdatt Dubhashi (Chalmers) were awarded a Research Environment Grant by the Swedish Research Council for the purposes of studying the extent to which African communities are trapped in poverty and how competing development interventions alter the communities’ prospects to free themselves from ‘poverty traps.’

About 300 million people in Africa live in extreme poverty. Given that living in impoverished communities can trap people in cycles of deprivation (‘poverty traps’), major development actors such as China and the World Bank have deployed a stream of projects to break these cycles (‘poverty targeting’). However, as scholars are held back by a data challenge, research has up until now been unable to answer fundamental questions such as whether poverty traps exist, and to evaluate to what extent interventions can release communities from such traps.

The aim of this project is to identify to what extent African communities are trapped in poverty and examine how competing development programs can alter these communities’ prospects to free themselves from deprivation.

To address this aim, we will (i) train image recognition algorithms—a form of AI—to identify local poverty from satellite images, 1984-2020; (ii) use these data to analyze how development actors affect African communities; (iii) use mixed methods to develop theories of the varieties of poverty traps; (iv), develop an R package, PovertyMachine, that will produce poverty estimates from new satellite images—ensuring that our innovations will benefit poverty research.

The research tasks are of such a challenging character that a single project or discipline cannot address it. Thus, seven social- and computer scientists have joined forces to tackle this project’s aim.

   

Funded By:

Acknowledgment:
This project is supported by the Combining Satellite images and Artificial Intelligence to Measure Poverty from 1982-2020 and Use These Data to Explain the Effects of World Bank and Chinese Development Programs in Africa grant (Swedish Research Council – 2020-0388), PIs: Adel Daoud (Primary); Maria Brandén; Devdatt Dubhashi; Peter Hedström; Fredrik Johansson; Ellen Lust; Xiao-Li Meng.