What do rental markets look like in the United States today? Over the course of three papers, I investigate rental housing markets in ten of America’s most populous cities: Boston, MA; Columbus, OH; Dallas, TX; Kansas City, MO; Minneapolis, MN; Nashville, TN; Omaha, NE; Philadelphia, PA; Seattle, WA; and Washington, DC. I compare the locations of landlords and rentals, examine the extent to which rental markets have concentrated ownership, and consider the difficult identification problem facing researchers who try to use administrative data to identify rental properties in the United States.
In the first paper, I geolocate rental properties and landlords in eight cities. I find that the median landlord has a mailing address within 10 miles of their rental property, and that a majority of landlords with a residential mailing address are located within the same region as their rental properties. Landlords with residential mailing addresses are located in neighborhoods that are whiter, richer, and have more college graduates than the neighborhoods in which they own properties. I also find that many landlords are located far away from their rental properties, in superstar cities and throughout the country. I use a network-science approach to identify the core locations of landlords, which I call the “landlord market area.”
In the second paper, I identify landlords who have significant market shares in a given city or neighborhood. I use machine learning to deduplicate different landlord records, a required step in the age of corporate landlords. I find that many neighborhoods have moderate and high levels of ownership concentration. Higher levels of concentration are correlated with higher rent levels among the cities I study. I use an instrumental variable approach to investigate the interaction between city-wide wage increases and ownership concentration, finding that neighborhoods with higher levels of concentration see larger rent increases.
In the third paper, I investigate the most common methods to identify rental properties in the United States. Most studies heretofore have relied on tax assessment databases to identify rental properties, yet I find that the most common approaches to do so are overinclusive of some types of units, and underinclusive of others. I compare these methods to the rental properties identified by rental registries and to American Community Survey estimates. I identify best practices with regard to rental registries, and interrogate when different approaches are more suitable than others.