Files
Abstract
Many societies are divided into multiple smaller groups. The defining feature of these groups is that certain kinds of interaction are more likely to take place within a group than across groups. I build a model in which group divisions are enforced through a reputational penalty for interacting with members of different groups. Agents who interact with members of different groups find that they can support lower levels of cooperation in the future. The model explains why agents may be punished by the other members of their group for interacting with members of different groups and why agents are punished for interacting with members of some groups but not others. I test the empirical implication that there should be less cooperation among members of groups that make up a larger percentage of their communities. I discuss the origin and possible future of social division.