Self-help groups and mutual assistance: Evidence from Urban Kenya.

Abstract This article examines the incomes of individuals who have joined self-help groups in poor neighborhoods of Nairobi. Self-help groups are often advocated as a way of facilitating income pooling. We find that incomes are indeed more correlated among individuals in the same group than among individuals who belong to different groups. Using an original methodology, we test whether this correlation is due to self-selection of similar individuals into the same groups. We find that this correlation is not driven by positive assortative matching. If anything, selection works in the opposite direction: incomes from group activities would be more correlated if individuals were matched at random. These findings are consistent with the idea that self-help groups play a mutual assistance role.