1 Residential Distribution in the City – Reexamined Itzhak Benenson, Ehud Or, Erez Hatna, Itzhak Omer Department of Geography and Human Environment, Tel-Aviv University Tel-Aviv, Israel bennya@post.tau.ac.il SUMMARY The recent boom in high-resolution GIS makes possible the investigation of urban residential distributions at the resolution of individual buildings and families. Availability of these data has inspired reexamination of the Schelling model of residential segregation and its application for simulating population patterns in real cities. The current paper argues that the Schelling model satisfies this criterion and consequently applies it to explain real-world residential pattern dynamic in nine Israeli cities. The study is based on data obtained from a unique Israeli census database in which individual and family records are geo-referenced to the layer of buildings. Analysis of the income-based residential patterns reveals their high heterogeneity – a mix of homo- and heterogeneous areas is typical for eight of the nine cities investigated. We explain this heterogeneity by the presence of a low fraction of wealthier householders who are highly tolerant of their poorer neighbors and can thus reside in their proximity. Extension of the model in this direction results in a qualitative correspondence between the model’s outcomes and the residential patterns observed in Israeli cities. KEYWORDS: Agent-Based modeling, Residential distribution, Population census, LOCAL INTERACTIONS – DO THEY DETERMINE URBAN RESIDENTIAL PATTERN? Population groups tend to segregate and sociological theory supports this view with the classic Schelling model of residential segregation considers individual agents, who care about the composition of their local neighborhood only. Basically, the model, proposed independently by Thomas Schelling (Schelling 1969) and James Sakoda (Sakoda 1971) considers social agent belonging to one of two mutually avoiding types – b and w, and relocating in reaction to the fraction of the agents of their own and of the strange types among the neighbors. To remind, the outcome of the Schelling model depends on the threshold fraction of strangers S Th the agent cannot tolerate, and on the initial population density. In case of characteristic of the city high population density, low sensitivity of the agents, say S Th ~ 80% and above, does not force agents to move and the residential pattern is preserved in time after short period of minor changes. The value of S Th between ~80% and ~20% results in segregated city (agents leave the neighborhoods with the high fraction of strangers and reside among friends), while high sensitivity to the strangers, S Th ~ 20% and below, results in unrealistic situation, when most of agents change their location each time step. Recent theoretical research reveals qualitative and robust character of the Schelling basic result. Namely, if the tendency to stay within those who are similar to you is not too weak, then the residential distribution in the city converges to segregation. The statement remains true for various definitions of the neighborhood, rules of relocation, irregular partition of space, etc. (Portugali and Benenson 1994; Portugali and Benenson 1995; Portugali and Benenson 1997; Benenson 1998; Benenson 1999). Recent experimental results confirm Schelling model outcome for the situation of the ethnic group competing for space (Benenson, Omer et al. 2002; Bruch and Mare 2004).