Grow the pie or have it? Using machine learning for impact heterogeneity in the Ultra-poor Graduation Model
Published Web Locationhttps://doi.org/10.26085/C3MG6Z
Anti-poverty interventions often face a trade-off between immediate reduction in poverty, measured by consumption, and building assets for longer-term gains. An “Ultra-poor Graduation” model, found effective on both dimensions, generally leans towards asset building. By using data from a large-scale RCT in Bangladesh, we find significant variation in impact on assets where the top quintile of gainers had an impact of 3.44 on their log of assets compared to the impact of 1.92 observed by the bottom quintile. We also find heterogeneity in household expenditure although the estimates are less robust across different estimation methods. Importantly, we find contrasts in characteristics of beneficiaries who made the most in assets vs. consumption. The results identify beneficiary characteristics that can be used in targeting households either to maximize impact on the desired dimension and/or to customize interventions for balancing the asset and consumption trade-off.