A Wealth and Status-Based Model of Residential Segregation

Abstract

We extend the classic “Schelling model” (1971 1978) to incorporate the wealth and status of agents and the desirability and affordability of residences. We analyze the effects of 1) the degree of the status-wealth correlation, and 2) the extent to which the wealth of residents shapes the affordability of residences, on levels of status and wealth segregation. Both factors generally exert a positive effect on both forms of segregation and interact to produce higher levels of segregation. The greater the correlation between status and wealth, the more the agents tend to segregate, either due to choice (for the wealthy and high status) or exclusion (for the poor and low status). We also find that housing price endogeneity is a precondition for status segregation.

Keywords:

• agent-based models
• residential choice
• Schelling
• segregation
• wealth
• status

We thank Mark Fossett, Ko Kuwabara, and Michael Macy for helpful comments and suggestions. We are also grateful for feedback from participants in the Artificial Societies seminar at Cornell and conference participants at Yale University and Hokkaido University.

Notes

¹But see Torrens and Nara (2005) for an exception.

²The model was written in NetLogo 3.1 (Wilensky, 1999).

³We also explored the effects of implementing an exponential wealth distribution and found nearly identical results.

⁴Subsequent analyses using the 1990 and 2000 censuses show that this conceptual scheme remains valid (Iceland, Weinberg, and Steinmetz 2002).

⁵ D has also been the subject of critique, with some researchers recommending alternative measures, particularly the variance ratio V (Coleman, Hoffer, and Kilgore, 1982). We also calculated V for all results presented here. We find that V and D are correlated with alpha > .99, and produce virtually identical substantive patterns. We therefore only report results for D.

⁶We also examined the results using a larger range of rounds and found virtually identical results, suggested the model had converged to a stable point by 500 rounds.

⁷This interaction effect, as well as the main effects of status-wealth correlation and price endogeneity, were also examined within a regression framework. We find positive and highly significant results for all three measures. Results are available upon request from the authors.

⁸Note that agents include their own cell in this search process. This prevents agents from moving to a cell that is more expensive than their own but cheaper than the cells on which they do not reside.