Retooling Poverty Targeting Using Out-of-Sample Validation and Machine Learning
Open accessAbstract
Proxy means test (PMT) poverty targeting tools have become common tools for beneficiary targeting and poverty assessment where full means tests are costly. Currently popular estimation procedures for generating these tools prioritize minimization of in-sample prediction errors; however, the objective in generating such tools is out-of-sample prediction. This paper presents evidence that prioritizing minimal out-of-sample error, identified through cross-validation and stochastic ensemble methods, in PMT tool development can substantially improve the out-of-sample performance of these targeting tools. The USAID poverty assessment tool and base data are used for demonstration of these methods; however, the methods applied in this paper should be considered for PMT and other poverty-targeting tool development more broadly.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The fundamental tradeoff between inclusion and exclusion error remains even with more sophisticated prediction methods.
“While more sophisticated prediction methods (like more flexible prediction equations, machine learning methods, and so on) can improve on these predictions (McBride and Nichols 2016), the fundamental tradeoff between inclusion and exclusion error that we document here remains.”
From Universal Basic Incomes versus Targeted Transfers: Anti-Poverty Programs in Developing Countries (2018) · cites this paper as McBride and Nichols 2016
The citation is attached to the assertion that more sophisticated methods can improve predictions; the sentence presents the inclusion–exclusion tradeoff as the citing authors’ own finding. The indexed passages provide repository metadata and an abstract, but no passage from the paper’s full text establishes the prediction-improvement assertion. Searches for machine-learning improvements and inclusion–exclusion error found no decisive full-text passage. (LLM, conf=0.72)
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cited by
Sorted by citations
Universal Basic Incomes versus Targeted Transfers: Anti-Poverty Programs in Developing Countries
Hanna, R., & Olken, B. A. (2018) · Journal of Economic Perspectives · doi:10.1257/jep.32.4.201 · Cited by 299 · see citing works
- Authors
- McBride, L., & Nichols, A.
- ORCID
- McBride, Linden
- Venue
- World Bank, Washington, DC eBooks
- Publisher
- World Bank, Washington, DC
- Published
- 2016-10
- OpenAlex
- W3122354151
- Cited by
- 12
- References
- 0
- Citations by year
- 2025: 3 · 2023: 1 · 2021: 1 · 2020: 3 · 2019: 3 · 2017: 1
- Type
- report
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- World Bank, Washington, DC eBooks (publishedVersion, cc-by)
- The World Bank Open Knowledge Repository (World Bank) (submittedVersion, cc-by)
- Topics
- Social Sciences › Safety Research › Poverty, Education, and Child Welfare
- Social Sciences › Sociology and Political Science › Income, Poverty, and Inequality
- Social Sciences › Decision Sciences › Management Science and Operations Research › demographic modeling and climate adaptation
- Keywords
- poverty targeting, proxy means test, out-of-sample validation, cross-validation, beneficiary selection, poverty measurement, machine learning
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
