Dynamic Properties of Poverty Targeting

A body of recent studies has compared the ability of proxy-means testing (PMT), a data-driven poverty targeting procedure, and community-based targeting (CBT), a participatory method,to identify consumption-poor households. Motivated by the facts that targeted benefits typically reach beneficiaries with a substantial time lag and that transitions into and out of poverty are frequent, we are first to assess PMT’s and CBT’s performance one and two years subsequent to the targeting exercise. With data from Burkina Faso, we replicate the finding that PMT targets more accurately than CBT with respect to poverty at baseline, by 14 percent. We find that this pattern is reversed for households’ poverty status twelve months later, while both methods perform identically with respect to poverty data collected 30 months after the baseline. We investigate how communities process different kinds of information and identify three properties of CBT that make it forward-looking: implicit weights put on PMT variables that predict future rather than current consumption, accounting for additional household characteristics not included in typical PMTs and processing of additional information unobserved by the researcher.