Models of social dynamics: Corruption, migration, and prejudice

This dissertation uses agent-based simulation models to shed new light on three different social dynamics that have proven difficult to study in other ways, because they involve feedback between micro-level behavior of heterogeneous individuals and macro-level patterns and outcomes, and because empirical findings are either very limited or appear conflicting. The three topics studied are: corruption, migration, and prejudice. Each chapter is a self-contained paper focused on one of these topics, containing both relevant background and theoretical analysis using a model of micro-foundations of the observed systemic behavior. The chapter on corruption shows how "tipping" from high to low systemic levels of corruption can occur endogenously, without any substantive change in system or structure. Tipping results from both limited information in individual decision-making and communication through social networks, which together allow perceptions of the risks of offering or receiving a bribe to spread rapidly. The migration chapter extends a previously developed model of the evolution of ethnocentrism, to show how migration in the context of ethnocentrism can lead to the formation of segregation and "balkanization" (even when the destination of migrants is random), and to show how migration can allow ethnocentrism to spread across societies and to rapidly socialize immigrants. The chapter on prejudice explores the potential for contact between ethnic groups to ameliorate in-group bias, offering a potential resolution of the "paradox" in empirical research on this topic by demonstrating a mechanism for the coexistence of both highly successful reciprocity between individual members of different groups and widespread ethnocentric in-group favoritism as groups come into contact. Movement, diversity, and bias in social influence are all shown to play important roles in determining the balance of these two effects. These chapters each individually generate insight about the micro-foundations of a specific topic. Together, they illustrate and extend the repertoire of agent-based modeling as a powerful research method for studying social dynamics.