Risk‐matching behavior in microcredit group formation: evidence from northern Ethiopia

Theoretical models on group lending assume the formation of groups of homogenous risk types. Recent theoretical and empirical findings challenge this view arguing that when markets for insurance are missing, risk homogeneity may not hold any more and risk heterogeneity can be the optimal outcome. Using data from an MFI in Tigray (Ethiopia), this article examines the homogeneity hypothesis and reflects on implications for repayment. No evidence is found that supports risk homogeneity, even accounting for matching frictions. However, we also do not find an explicit link between the presence of risk heterogeneity and side‐payments due to missing insurance as suggested in the literature. Instead, other trust‐based social networks seem to underlie heterogeneity. Such social networks are often synchronized with credit groups and influence the probability of repayment under heterogeneity. The implication is that successful repayment rates in group lending need not arise only under risk homogeneity but can also arise under risk heterogeneity. Heterogeneity may also serve to bridge missing insurance markets in poor rural environments. MFIs therefore need to consider such local conditions when designing their lending schemes.