Exploring the dynamics of neighbourhood ethnic segregation with agent-based modelling: an empirical application to Bradford, UK

How individuals’ residential moves translate into overall emergent segregation patterns remains a key challenge in neighbourhood ethnic segregation research. In this paper, we use agent-based modelling to explore this concern, focusing on the interactive role of ethnic and socio-economic homophilic preferences and socioeconomic housing constraints as determinants of residential choice. Specifically, we extend the classic Schelling model to a random utility discrete choice approach to simulate the relocation decisions of people. We model different weights for preferences for ethnic and socioeconomic similarity in neighbourhood composition over random relocations, in addition to housing constraints. We formalise how different combinations of these variables could replicate empirically observed ethnic segregation scenarios in Bradford, a substantially segregated local authority in the UK. We initialise our model with geo-referenced data from the 2011 Census and use various measures of segregation to describe our results. As in the original Schelling model, we find that even mild ethnic preferences alone would lead to unrealistic ethnic over-segregation in Bradford. However, we demonstrate that such process can be altered in favour of less ethnic segregation when agents’ preferences for socioeconomic similarity are slightly stronger than their preferences for ethnic similarity. We discuss theoretical and policy contributions of our findings.