Racial covenants, which were used throughout the U.S. during the first half of the 20th century, were clauses in property deeds that prohibited the sale or rental of property to racial and ethnic minorities. This paper studies the long-run causal and persistent effects of racially restrictive covenants on racial sorting, racial homeownership differentials, and house prices. Using novel data on racial covenants and a quasi-experimental design that exploits time to build delays in housing and the 1948 U.S. Supreme Court ruling that made racial covenants unenforceable, we compare newly built, covenanted (and hence all-white) neighborhoods built before 1948 with covenanted neighborhoods constructed after the verdict, as the post-1948 neighborhoods lost a primary legal way to coordinate segregation. We find that about 5-17% of the observed neighborhood racial residential sorting and 6-17% of racial homeownership sorting, measured from 1980 to 2020, can be causally linked to racial covenants of the past. In addition, while the treated and control neighborhoods exhibit some differences in public amenities, particularly in the distance to highways and the restrictiveness of zoning regulations, these differences do not drive the observed persistent effects. Instead, homophily bias or unobserved housing quality are the likely drivers behind the persistent effects.
--Dan Ernst
