Hidden Biases: Selective Advertising in the Rental Housing Market

A growing body of literature has documented persistent racial discrimination in rental housing markets, often using experimental (audit or correspondence) studies where fictitious identities request to view a housing unit. These methods are not designed to capture a subtle form of discrimination where landlords choose not to advertise certain units, instead reserving them for prospective tenants after an initial screening or other process to select renters on certain attributes. These selectively unadvertised units will not appear in the sample for an experimental analysis, potentially biasing the results of these studies. In this paper, we introduce an innovative method for detecting selective advertising. Our approach uses a large marketing dataset to track property units turnovers in 27 major U.S. metropolitan areas. We match this data with a rental listings dataset to identify turnover units that are not publicly advertised. By comparing the racial composition of occupants in listed versus “hidden” units and controlling for alternative factors that might account for racial sorting, we assess the extent of discrimination through selective advertising. We find that this form of discrimination against Black and Hispanic renters is particularly severe in neighborhoods with better amenities, where other forms of discrimination are more restricted, and in neighborhoods nearing “tipping points” in racial composition.