Description: Labor-Public Workshop: Heath Henderson (Drake University)

Location: 368A Heady Hall

Contact Person: John Winters

Title: Poverty Mapping in the Age of Machine Learning

Abstract: Recent years have witnessed considerable methodological advances in poverty mapping, much of which has focused on the application of modern machine-learning approaches to remotely-sensed data. Poverty maps produced with these methods generally share a common validation procedure, which assesses model performance by comparing sub-national poverty estimates with survey-based, direct estimates. While unbiased, direct estimates can be imprecise measures of true poverty rates, meaning that it is unclear whether the standard validation procedures are informative of actual model performance. In this paper, we examine the credibility of existing approaches to model validation by constructing a pseudo-census from the Mexican Intercensal Survey of 2015, which we use to conduct several simulation experiments. We find that the standard validation procedure can be misleading in terms of model assessment, with notable implications for model selection. Further, we find that our closest approximation to existing machine-learning approaches performs poorly when evaluated against “true” poverty rates and fails to outperform traditional poverty mapping methods in targeting simulations.