Empirical research on the impact and determinants of group lending is by now substantial. However, very little is known about the possible role of collateral to mitigate incentive problems in group lending. This is because microcredit programs have normally been implemented in rural areas of developing countries. Indeed, the reason for this choice is lack of credit access since agents with collateral are very rare. Also, to the extent that rural communities have tight-knit hierarchical structures information about borrowers is accessible and the enforcement of sanctions via social networks makes collateral superfluous for default mitigation. Yet, in an urban setting in which information is more atomized and social sanctions are not as powerful, collateral may have an important role in group lending. First, we illustrate in a model the role of collateral to mitigate group default. Second, we use data from a group lending program implemented in 2001 in Cotonou, the largest city in Benin with more than one million inhabitants. We empirically explore the risk profile of individual borrowers and resulting group heterogeneity to identify the role of personal contributions to investment projects. Our evidence suggests that while diversification within groups facilitates risk pooling, it also increases expected bailout or group default costs for low risk borrowers. Collateral helps offset and alleviate potential negative spillovers from group default induced by membership of borrowers with risky projects. The presence of borrowers with collateral facilitates access to credit for group members without collateral, who in turn provide insurance against group default. We find joint liability to be a mechanism for risk sharing in a setting where poor households lack resources for collateral and insurance markets are missing.