Segregating socioeconomic classes leads to an unequal redistribution of wealth

The research question explores how socioeconomic class segregation and the use of punishment influence cooperation and wealth distribution, particularly in societies with high income inequality. The study's context is rooted in the observation that animal and human societies are structured hierarchically, often exhibiting significant socioeconomic inequalities. These inequalities, deeply ingrained in social structures, have been a subject of scholarly interest for decades, with explanations ranging from the shift to agriculture and industrialization to the inherent nature of complex social interactions in larger groups. The researchers hypothesize that unequal resource holdings affect the willingness to cooperate, particularly in social dilemmas, influencing both current interactions and future expectations. A lack of reciprocity between richer and poorer classes can further hinder cooperation, reducing civic engagement and negatively impacting economic growth. The study highlights that socioeconomic class segregation is pronounced in societies with high levels of income inequality, making it a crucial factor to investigate in understanding cooperation and wealth distribution. The researchers aim to model this socioeconomic class segregation and analyze how it affects the behavior and earnings of individuals using the Iterated Prisoner's Dilemma Game (IPDG), with a specific focus on China, a country characterized by significant income inequality. The study seeks to empirically demonstrate how allowing for interaction and punishment across socioeconomic classes, rather than maintaining segregation, may lead to a more equitable redistribution of wealth.

The paper reviews existing literature on social hierarchies, cooperation, and punishment in both animal and human societies. It cites works examining the historical evolution of social hierarchies, the role of asymmetric power in shaping social relationships, and the impact of unequal resource holdings on cooperation. The authors refer to studies illustrating how inequalities affect willingness to cooperate in social dilemmas, impacting both current interactions and future expectations. Research on the lack of reciprocity between socioeconomic classes and its effect on civic engagement and economic growth is also reviewed. The literature on socioeconomic class segregation in relation to income inequality is presented, alongside studies on how modeling the behavior of individuals from such socioeconomic structures contributes to a better understanding of cooperation's impact at an evolutionary level. The review also incorporates research on repeated social interactions, focusing on punishment as a strategy and its implications in scenarios of asymmetric interests. Finally, the paper acknowledges previous work on the Prisoner's Dilemma Game and the impact of punishment power on players' payoffs, setting the stage for its own experimental design that incorporates these elements within the context of socioeconomic segregation in China.

The study employed a combination of laboratory experiments and agent-based modeling. The laboratory experiments were conducted at Yunnan University of Finance and Economics in Kunming, China, with 348 students (164 males, 184 females) aged 18–22, from various regions of China. Participants, unfamiliar with game theory and behavioral economics, provided informed consent. The experiments, approved by the university's ethics committee, involved three treatments: an integrated condition (T0) and two segregated conditions (T1 and T2). In the integrated condition, participants were randomly paired for an Iterated Prisoner's Dilemma Game (IPDG) with a costly punishment option. In the segregated conditions, participants were paired based on their socioeconomic status, with either the richer (T1) or poorer (T2) group having the ability to punish the other. A socioeconomic index was created based on a questionnaire assessing participants' family income and assets. This index divided participants into richer and poorer classes. The IPDG allowed choices between cooperation, defection, and punishment (CDP) or only cooperation and defection (CD). The experiments involved repeated interactions, with payoffs awarded based on performance. Following the experiments, an agent-based model was developed to simulate the IPDG with parameters mirroring the human experiments, allowing for comparisons between human and artificial agent behavior and for testing various hypotheses about cooperation, defection, and punishment under different conditions. Data were analyzed using generalized linear mixed models (GLMMs).

The experimental results revealed significant differences in behavior and earnings across treatments. In the integrated condition (T0), cooperation levels were similar across socioeconomic classes, indicating that the ability to punish each other restrained defection. However, in the segregated conditions (T1 and T2), the punishing classes decreased their cooperation while punishing, leading to a significant difference in earnings. The punishing classes earned almost double the amount of the non-punishing classes. Conversely, in the integrated condition, earnings differences were statistically insignificant. Agent-based simulations confirmed these findings, showing similar patterns of behavior and earnings distribution. When the agent-based model mirrored the conditions of the human experiments (using a short-term memory strategy and a predisposition toward defection), it reproduced the human results remarkably well. The model also suggested that the strategy employed by subjects relied on the memory of the previous five rounds, implying a tendency towards defection in initial interactions. In summary, the study showed that socioeconomic segregation led to increased inequality, whereas integration resulted in a more equitable distribution of wealth.

The findings directly address the research question by demonstrating the significant impact of socioeconomic segregation on cooperation and wealth distribution. The disparity in earnings between the punishing and non-punishing classes in the segregated conditions highlights how such segregation exacerbates inequality. In contrast, the lack of significant earnings differences in the integrated condition supports the hypothesis that allowing for cross-class interactions and reciprocal punishment can lead to a more equitable outcome. The close correspondence between the human experiment results and the agent-based simulations further strengthens the findings' validity, suggesting that the observed patterns are not merely artifacts of the experimental design but reflect underlying behavioral mechanisms. The prevalence of defection in initial interactions, particularly in the Chinese context, may be attributed to lower trust levels compared to Western societies. These findings contribute significantly to the understanding of how social structures influence economic outcomes and highlight the potential of policy interventions promoting social integration to address income inequality.

This study demonstrates that socioeconomic segregation significantly increases wealth inequality, while integration promotes a more equitable distribution of resources. The use of an Iterated Prisoner's Dilemma Game and agent-based modeling provides strong evidence supporting this conclusion. Future research could explore the long-term effects of social integration policies and investigate the influence of cultural norms and trust levels on cooperation in different societal contexts. Further research could also explore the impact of different punishment mechanisms and varying levels of income inequality on cooperation and wealth distribution.

The study's sample was limited to university students in China. The socioeconomic index used to classify participants, while comprehensive, may not capture the full complexity of socioeconomic status. The experimental setting, though carefully designed, may not perfectly replicate real-world social interactions. The study's focus on a specific type of social dilemma might limit the generalizability of the findings to other types of economic interactions.