The paper by Michihiro Kandori examines how social norms and community enforcement can sustain cooperation in repeated interactions, especially when agents change partners over time. It extends the theory of self-enforcing agreements in repeated games to situations where agents may not have long-term relationships. The key idea is that cooperation can be maintained through social norms that rely on self-interested agents' incentives, even when they have limited information about others' actions. The paper presents two main results. The first shows that a community can sustain cooperation even when each agent only observes the results of the trades they personally engage in. The second result provides a Folk Theorem under mild assumptions, showing that any mutually beneficial outcome can be sustained when agents carry labels (such as reputation, membership, or license) that are systematically revised. The paper discusses the role of information in repeated matching games, emphasizing that the effectiveness of community enforcement depends on the transmission of information among agents. It also highlights the importance of local information processing, where agents use labels to update their behavior based on past interactions. This mechanism allows for the maintenance of efficient outcomes even in the absence of perfect information. The paper also addresses the limitations of social norms under no information processing, where cooperation is difficult to sustain due to the lack of effective punishment mechanisms. It argues that systematic information processing is essential for maintaining social norms and achieving efficient outcomes in economic transactions. The study concludes that community enforcement can be as effective as personal enforcement in repeated games, especially when information is transmitted systematically.The paper by Michihiro Kandori examines how social norms and community enforcement can sustain cooperation in repeated interactions, especially when agents change partners over time. It extends the theory of self-enforcing agreements in repeated games to situations where agents may not have long-term relationships. The key idea is that cooperation can be maintained through social norms that rely on self-interested agents' incentives, even when they have limited information about others' actions. The paper presents two main results. The first shows that a community can sustain cooperation even when each agent only observes the results of the trades they personally engage in. The second result provides a Folk Theorem under mild assumptions, showing that any mutually beneficial outcome can be sustained when agents carry labels (such as reputation, membership, or license) that are systematically revised. The paper discusses the role of information in repeated matching games, emphasizing that the effectiveness of community enforcement depends on the transmission of information among agents. It also highlights the importance of local information processing, where agents use labels to update their behavior based on past interactions. This mechanism allows for the maintenance of efficient outcomes even in the absence of perfect information. The paper also addresses the limitations of social norms under no information processing, where cooperation is difficult to sustain due to the lack of effective punishment mechanisms. It argues that systematic information processing is essential for maintaining social norms and achieving efficient outcomes in economic transactions. The study concludes that community enforcement can be as effective as personal enforcement in repeated games, especially when information is transmitted systematically.