Racial discrimination and segregation in housing have been important aspects of reality in the United States over the entire history of the country.  Individual preferences concerning the racial composition of one’s neighborhood has led through private and collective actions to high levels of racial segregation.  Although aversion on the part of white owners of rental housing to dealing with blacks might have led to differences in the rents that blacks and whites pay for identical housing in identical neighborhoods, the results of the best empirical studies on this question are mixed.  Some find that blacks pay a premium; others find that they receive a discount; and still others find no difference.  The reason for the mixed results is almost surely that the data underlying different studies is deficient in different ways.

The purpose of this paper is to estimate the difference in the rents that blacks and whites pay for identical housing in identical neighborhoods and the differences in the rents that each group pays for otherwise identical units in neighborhoods with different racial compositions.  A key advantage of this study of racial differences is the use of an unusually rich data set to compute housing rental gaps between whites and African-Americans in the U.S.  This data set overcomes the shortcomings of the data sets used in all previous studies.  It is large, containing over 400,000 observations, covers all parts of the United States, contains detailed information about the housing units and their immediate neighborhoods, and importantly, given the sample size, there are many blacks and whites in neighborhoods with each racial composition.

Results suggest that households lead by African Americans pay on average 3 percent more for housing than their white counterparts.  This rent gap seems to increase with the share of white population in the census tract and decrease with the neighborhood’s poverty rate.  For example, while African-American households pay roughly 2 percent more than whites when occupying a housing unit in neighborhoods with less than 10 percent of the population white, the premium increases to about 4 percent in racially integrated neighborhoods (40-60 percent of the population white).  Another benefit of the data is the ability to produce estimates of rental gaps separately for 376 distinct metropolitan areas or the nonmetropolitan parts of states.  We find significant variation in racial rent gaps across geographic areas and that those differences are correlated with Census region.