liu.seSearch for publications in DiVA
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Using Machine Learning and Daytime Satellite Imagery to Estimate Aid's Effect on Wealth: Comparing China and World Bank Programs in Africa
Linköping University, Department of Management and Engineering, The Institute for Analytical Sociology, IAS. (IAS)
2024 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

A large literature has not reached consensus on foreign aid’s economic effects. Using geolocated aid data and daytime satellite images over nearly 10,000 African neighborhoods, I examine the economic growth impact of World Bank and Chinese aid to 36 Africa countries from 2002-2013, covering 88% of the continent’s population, by sector (e.g. Health, Education, Water Supply and Sanitation, etc.).  I estimate each funder and aid sector’s average treatment effect with an inverse probability weighting approach and adjust for two types of confounders: those I provide in a tabular format and proxies based on satellite images of each neighborhood. The use of image-based confounders may reduce bias due to omitted variables and measurement errors when unobserved or mis-measured variables are visible remotely.  To measure economic outcomes, I use a new wealth index generated by a machine learning algorithm trained to associate USAID-funded DHS survey wealth measures with daytime and nighttime satellite imagery from the same years and locations. The availability of the wealth estimate for 3-year periods over thirty years enabled the analysis to use panel data and fixed effects at the second administrative division (e.g. county, district, city) level. The results are heterogenous across sectors but generally show small positive effects of World Bank aid and larger positive effects of Chinese aid.  Substantive results are generally robust to the choice of computer vision image model, except for three funder-sectors where wide confidence intervals make one model but not the other statistically insignificant.

Place, publisher, year, edition, pages
2024. , p. 291
Keywords [en]
machine learning, daytime satellite imagery, causal inference, foreign aid, poverty, world bank, china
National Category
Social Sciences Interdisciplinary
Identifiers
URN: urn:nbn:se:liu:diva-205256ISRN: LIU-IEI-FIL-A--24/04656--SEOAI: oai:DiVA.org:liu-205256DiVA, id: diva2:1875097
Subject / course
Master’s Thesis in Computational Social Science
Presentation
2024-06-05, 16:34 (English)
Supervisors
Examiners
Available from: 2024-06-26 Created: 2024-06-20 Last updated: 2024-06-26Bibliographically approved

Open Access in DiVA

ConlinThesisAndAppendices(21667 kB)40 downloads
File information
File name FULLTEXT01.pdfFile size 21667 kBChecksum SHA-512
accea90a39f072e97952e52de01b207c2232175341053b8dc38f95ff5ce6dcf6557a6af86d726d4540dde0ed8543c2bfb76a149ee642e5dd5f7d9f509ab2cba0
Type fulltextMimetype application/pdf

Search in DiVA

By author/editor
Conlin, Cindy
By organisation
The Institute for Analytical Sociology, IAS
Social Sciences Interdisciplinary

Search outside of DiVA

GoogleGoogle Scholar
Total: 40 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 228 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf