Decentralized versus Statistical Targeting of Anti-Poverty Programs: Evidence from Burkina Faso

Targeting of national anti-poverty programs in low-income countries commonly relies on statistical procedures involving household-level survey data, while small-scale poverty-alleviation programs often employ so-called community-based targeting, where village communities themselves identify program beneficiaries. Combining data from community-based targeting exercises in north-western Burkina Faso with household-level survey data, we compare the targeting accuracy of community-based targeting with several statistical procedures when the program's purpose is to target consumption-poor households. We find that the community-based assessment targets a similar share of consumption-poor households as the best-performing statistical procedures which are not calibrated with household-level consumption data. Community-based targeting performs relatively better in urban than in rural areas and is not at a disadvantage in larger or more heterogeneous communities. In a cost-benefit analysis we find that in our sub-Saharan African context community-based targeting is far more cost-effective than any statistical procedure for common amounts of welfare program benefits.