Association for Public Policy Analysis & Management

Privately owned and operated rental housing has increasingly become the mechanism through which housing subsidies are distributed to poor families, specifically the Housing Choice Voucher Program which currently serves approximately 2.4 million families. Countless papers have examined the impact of utilizing private owners on the neighborhood attainment of poor families, focusing on neighborhood quality and residential stability. While invaluable, these studies have historically overlooked the fact that voucher mobility patterns are shaped by two sets of actors – subsidized tenants and private landlords – who respond to one another dynamically to shape the geography of opportunity in America’s cities.

In recent qualitative work, we drew on over 160 interviews with private landlords to examine the process by which they chose to participate in the HCV program or refused to do so (Garboden et al. 2018, Garboden and Rosen 2018). Perhaps most surprisingly was our finding that over 70 percent of the non-participating landlords (with units at or near FMR) had previously accepted vouchers but had exited the program after a negative experience. Our set of landlords was sampled randomly from rental listing, but was fairly small and thus did not accommodate precise point estimates or stratification by neighborhood condition.

This current paper uses administrative data from the Department of Housing and Urban Development (HUD) in four cities (Baltimore, Cleveland, Dallas, Washington, DC) to examine this phenomenon directly. Of primary interest is the degree to which properties that are subsidized in a particular year remain in the voucher program in subsequent years. We are particularly interested in what neighborhood attributes (both cross sectional and trend) are predictive of supply-side exit from or entrance into the HCV program, either with or without the sale of the property. We use the HUD 50058 administrative dataset for our four study site locations. This restricted-use dataset provides annual address-level data on all HCV-subsidized households between 1994 and 2015 in the Baltimore, Dallas, Cleveland, and Washington, DC metropolitan areas and includes both tenant demographics and an owner unique identifier. This data will allow us to consider the role of tenant turnover in this process. National sales data supplied via an MOU with Zillow.com allow us to examine whether supply-side program exits are associated with unit sale.