Urban Networks and Targeting: Evidence from Liberia

To solicit information from a social network in urban Monrovia, Liberia, we ask community members to assess their neighbors' poverty and assist in targeting a cash transfer. We find little evidence that local leaders, randomly selected neighbors, or neighbors nominated by fellow community members can accurately assess whether others in their community are among the poorest. Yet, all three groups target transfers to the poor modestly better than would be attributable to chance but worse than proxy-means-test-based targeting. We conclude that community advice provides some information for targeting but there are significant information frictions in networks in Monrovia.