Houston has no zoning, and that fact has made it the standing natural experiment in land-use scholarship: compare Houston with a zoned peer, and whatever differs is attributed to zoning. Christopher Berry’s widely cited comparison of Houston and Dallas (2001) found no significant difference in Black–White segregation and concluded zoning was unlikely to be a major driver of it.
The comparison is pitched at the wrong scale. Zoning is a municipal instrument, and its effect operates through competition among jurisdictions within a metropolis, the sorting the Tiebout mechanism describes, not through differences between metropolises, which average the mechanism away. Re-partitioning each region into principal city versus suburbs and testing within each also reveals the two cities as poor partners for the comparison: Houston’s principal city holds two-thirds of its region’s population, Dallas’s under a quarter.
Measured with the spatial information theory index, chosen because an additive decomposition across the city–suburb partition is built into it, and inferred by comparative spatial inference with 10,000 permutations per comparison, the analysis re-runs the original on 1990 census and 2021 ACS tracts, Black–White throughout. Between metropolises, the original’s null result holds in both years. Inside each region, the city–suburb test separates them: in 1990 both principal cities were significantly more segregated than their suburbs; by 2021 the city of Dallas remains so, and the city of Houston no longer is. Three decades of the same macroeconomic forces, two land-use regimes. The analysis runs in the lab’s segregation package.