Social network structure and the radius of risk sharing

Abstract Informal risk sharing, an important coping strategy, is mediated by social networks. Does risk sharing extend beyond immediate connections? If so, what is its radius? I examine the radius of risk sharing as a function of network structure. To do so, I employ community detection—a tool imported from network science—and dyadic regression. I find evidence the radius of risk sharing extends beyond direct connections. Using data from a behavioral experiment in Colombia, I find that detected community co-membership and distance-2 connections (i.e., friends of friends) explain co-membership in experimental risk sharing groups. Using data from a village census in Tanzania, I find that distance-2 and 3 connections explain risk sharing transfers, but detected community co-membership does not. I address a crucial issue of network sampling in the Colombia illustration using simulation methods, finding my preferred specification is robust to this concern. These methods may benefit those who seek to understand the quality of risk sharing when risk sharing groups are loosely defined and illegible to outsiders.