Schelling's segregation model: Parameters, scaling, and aggregation

Abhinav Singh, Dmitri Vainchtein, Howard Weiss

Research output: Contribution to journalArticlepeer-review

25 Scopus citations


Thomas Schelling proposed a simple spatial model to illustrate how, even with relatively mild assumptions on each individual's nearest neighbor preferences, an integtated city would likely unravel to a segregated city, even if all individuals prefer integration. This agent based lattice model has become quite influential amongst social scientists, demographers, and economists. Aggregation relates to individuals coming together to form groups and Schelling equated global aggregation with segregation. Many authors assumed that the segregation which Schelling observed in simulations on very small cities persists for larger, realistic sized cities. We describe how different measures can be used to quantify the segregation and unlock its dependence on city size, disparate neighbor comfortability threshold, and population density. We develop highly efficient simulation algorithms and quantify aggregation in large cities based on thousands of trials. We identify distinct scales of global aggregation. In particular, we show that for the values of disparate neighbor comfortability threshold used by Schelling, the striking global aggregation Schelling observed is strictly a small city phenomenon. We also discover several scaling laws for the aggregation measures. Along the way we prove that in the Schelling model, in the process of evolution, the total perimeter of the interface between the different agents always decreases, which provides a useful analytical tool to study the evolution.

Original languageEnglish (US)
Pages (from-to)341-366
Number of pages26
JournalDemographic Research
StatePublished - Sep 15 2009

All Science Journal Classification (ASJC) codes

  • Demography


Dive into the research topics of 'Schelling's segregation model: Parameters, scaling, and aggregation'. Together they form a unique fingerprint.

Cite this