Robust estimates of vulnerability to poverty using quantile models

Standard methodologies that are used to identify households vulnerable to future episodes of poverty rely on distributional assumptions that may lead to classification errors. This paper shows how quantile models improve the identification of vulnerable households by relaxing these distributional assumptions. Quantile models are robust to outliers and classical measurement error, and easy to implement which allows easy adoption by policymakers. Applying this quantile strategy to data from Uganda to illustrate its superiority to standard approaches, I find that it more accurately identifies the future poor among the general population. The accuracy is highlighted in the fact that more than 2 in 3 households identified as vulnerable using the quantile strategy became/remained poor within 1–2 years, compared with less than 1 in 2 households using standard empirical strategies. Overall, this study points to gains that researchers and policy practitioners can make by relaxing distributional assumptions when identifying the vulnerable.