Spatial Variations in Exclusionary Criteria from Online Rental Advertisements

Abstract

Although the Fair Housing Act prohibits discrimination in the rental housing market on the basis of race, color, or religion, there exist ample opportunities for landlords to restrict their rental units to individuals with various backgrounds using exclusionary screening practices. These criteria, such as minimum credit scores, criminal history, source of income, or prior evictions, for instance, are often correlated with race and thus hold the potential to perpetuate spatial patterns of racial and income segregation. In this article, we analyzed online rental listings from Zillow and Craigslist in a case study of Charlotte, North Carolina, to examine the proliferation and spatial variation in exclusionary criteria by neighborhood race and income. Overall, we found criminal background, credit score, housing voucher, eviction, and minimum income restrictions to be greater in poorer and minority neighborhoods. When distinguishing by platform, however, we found the presence of corporate landlords, who did not advertise on Craigslist, to be a significant driver of these results. Corporate landlords represented nearly 60 percent of advertisements on Zillow and were more common in poorer and minority neighborhoods; they also systematically included nearly all restrictions. Craigslist, by contrast, had fewer criteria that correlated with race and income. Key Words: housing discrimination, natural language processing, online rental listings, spatial analysis, Web scraping.