Forecasts in Schelling's segregation model

Abstract

In Schelling's segregation model [9], the successive moves of agents optimizing their own locations lead to a suboptimal segregated distribution of the population, even though all agents have the same preference for mixed neighborhoods. One of the main assumptions underlying this general result of segregation models is that agents rely on comparisons between instantaneous utilities in order to make their moving decisions. On the contrary and certainly more reasonably, we assume in this article that agents forecast later states using a linear extrapolation of past states heuristic in order to make their decisions. We show that for a relatively small set of parameters, considering forecasting agents allows to dramatically reduce sub-optimality in a framework close to [9]'s model.

Avalanches in an extended Schelling model: an explanation of urban gentrification

“…Further studies should focus on variants in which agents consider their future happiness perspectives [25], or the influence of altruistic behaviour [26] and the balance between cooperative and individual dynamics [27].…”