Housing Discrimination among Available Housing Units in 2012: Do Paired Testing Studies understate Housing Discrimination?

In 2012, the Department of Housing and Urban Development launched the fourth major nationwide housing discrimination study with the goal of measuring housing discrimination in rental and owner-occupied housing for blacks, Hispanics and Asians. The substantial declines in discrimination observed between 1989 and 2000 lead to a considerable debate about whether paired testing studies of the type conducted in 2000 understate the extent of housing discrimination. Using the 2012 study and earlier evidence, this paper considers three of the significant concerns about paired testing studies of housing discrimination: 1. exclusion of minorities during the process of setting up appointments, 2. the net measure of adverse treatment understating discrimination because some housing units are systematically not shown to white testers, and 3. the use of metropolitan wide advertisements for conducting tests may systematically miss neighborhoods where discrimination is higher. The 2012 study directly addresses the first concern finding at most very low levels of discrimination in obtaining an appointment over the phone. The evidence for the second concern is mixed. The levels of equal treatment in 2012 in terms of basic access were quite high leaving little room for the systematic exclusion of white under some circumstances, while steering persisted against both blacks and Asians in owner-occupied housing. Nonetheless, estimates of steering were modest, and the gross levels of adverse treatment are far too large to be explained by the steering observed in 2012. In order to address the third concern, this paper conducts a new empirical analysis in which we measure the availability of rental and owner-occupied housing in each broad neighborhood represented in the 2012 Housing Discrimination Study, and reweight the tests to represent the spatial availability of housing across each metropolitan site rather than the distribution of advertisements. While the reweighting substantially changed the weights on individual tests, the average attributes of the neighborhoods represented by those tests experienced only modest changes from reweighting, and the estimated measures of adverse treatment were unchanged.