Urban Segregation as a Complex System: An Agent-Based Simulation Approach


Flávia F. Feitosa (flavia@dpi.inpe.br)


Urban segregation represents a significant barrier for achieving social inclusion in cities. To overcome this, it is necessary to implement policies founded upon a better understanding of segregation dynamics. However, a crucial challenge for achieving such understanding relies on the fact that segregation is a complex system. It emerges from local interactions able to produce unexpected and counterintuitive outcomes that cannot be defined a priori.

This study adopts an agent-based simulation approach that addresses the complex nature of segregation. It proposes a model named MASUS, Multi-Agent Simulator for Urban Segregation, which provides a virtual laboratory for exploring theoretical issues and policy approaches concerning segregation. The MASUS model was first implemented for São José dos Campos, a medium-sized Brazilian city. Based on the data of this city, the model was parameterized and calibrated.

The MASUS’ potentiality is demonstrated through three different sets of simulation experiments: the first compares simulated data with real data, the second tests theories about segregation, and the third explores the impact of anti-segregation policies. The first set of experiments provides a retrospective validation of the MASUS model by simulating the segregation dynamics of São José dos Campos during the period 1991-2000. In general, simulated and real data reveal the same trends, a result that demonstrates that the model is able to accurately represent the segregation dynamics of the study area.

The second set of experiments aims at demonstrating the potential of the MASUS model to explore and test theoretical issues about urban segregation. These experiments explore the impact of two mechanisms on segregation: income inequality and personal preferences. To test the impact of income inequality, scenarios considering different income distributions were simulated and compared. The results show how decreasing levels of income inequality promote the spatial integration of different social groups in the city. Additional tests were conducted to explore how the preferences of high-income families regarding the presence of other income groups could affect segregation patterns. The results reveal that the high levels of segregation were maintained even in a scenario where affluent households did not take into account the income composition of neighborhoods when selecting their residential location.

Finally, the third set of experiments provides new insights about the impact of different urban policies on segregation. One experiment tests whether the regularization of clandestine settlements and equitable distribution of infrastructure would affect the segregation trends in the city. The simulated outputs indicated that they had no significant impact on the segregation patterns. Besides this test focusing on a general urban policy, two specific social-mix policy approaches were explored: poverty dispersion and wealth dispersion. The result suggests that policies based on poverty dispersion, which have been adopted in cities in Europe and the United States, are less effective in developing countries, where poor families represent a large share of the population. On the other hand, the policy based on wealth dispersion was able to produce substantial and long-term improvements in the segregation patterns of the city.

Back to Publications

QR Code
QR Code wiki:ourpub:segregation_complex (generated for current page)