Landlord Aversion and Residential Segregation: An Agent-Based Simulation of Micro Segregation

• In Event: 50205 - Big Problems, Middle-Range Theories

Abstract

Housing agents and landlords play a crucial role in shaping residential segregation. However, the extent to which their behaviors influence segregation at multiple scales—such as the building and neighborhood levels—remains unclear. This paper examines how landlord aversion to minority residents affects residential segregation using a novel agent-based modeling (ABM) approach. Our computational simulation captures segregation dynamics within a more realistic metropolitan landscape, where neighborhoods consist of both single-unit and multi-unit properties owned or managed by different landlords. We simulate three major scenarios by varying resident preferences, landlord aversion, and resident socioeconomic resources to assess their impacts on segregation at both the neighborhood and landlord levels. Preliminary findings indicate that landlord aversion significantly contributes to heightened micro-segregation at the landlord level. Minority residents, constrained by landlords’ preferences, tend to relocate to properties owned by landlords who are more accepting of minority tenants. Depending on the spatial distribution of properties owned by landlords with different attitudes, this clustering effect, driven by landlord aversion, can reduce segregation as measured by the conventional dissimilarity index, which primarily captures neighborhood-level segregation. Our ABM reveals that while minority residents may appear spatially dispersed throughout the city, they remain segregated at the property level based on landlord ownership. These findings provide empirical support for the role of landlords and housing agents as key mechanisms in shaping segregation patterns, offering a new theoretical explanation for the observed decline in neighborhood-level segregation in urban centers over recent decades. Methodologically, we enhance existing ABM approaches to residential segregation by expanding traditional two-dimensional analyses into a multidimensional framework.

Authors

• Hao Liang, Cornell University

• Keiichi Satoh, Hitotsubashi University