The Costs of Environmental Land-Use Regulation

This paper examines how environmental land-use regulation under the ESA affects housing construction. Species protections offer an unusually useful research setting because their regulatory force varies across both time and geography: A listing imposes new restrictions only in places that overlap the species’ habitat. I combined Census Bureau data on housing permits with US Fish and Wildlife Service records identifying the timing, location, and characteristics of species listings. The principal sample contains a balanced panel of 12,664 permit-issuing places from 1980 through 2024 and covers roughly two-thirds of all US housing permits.

To estimate the effects of regulation, I compared changes in permitting within places covered by a newly listed species’ habitat with changes in places not covered by a listing at that time. The analysis combines 661 animal-listing events in a stacked difference-in-differences design, using an eight-year period before and after each listing. I drew comparisons among places in the same state or metropolitan area and incorporated matching and reweighting procedures that make treated and untreated places more similar before a listing. The resulting sample contains approximately 4.7 million place-event-year observations.

The findings indicate that an additional endangered species listing reduced annual housing permits by approximately 0.52 units per 1,000 residents. This represents about 9–10 percent of the average annual permitting rate. Permit trends in treated and comparison places were similar before listings took effect, but permitting in treated places fell relative to nearby controls afterward. The result remained negative across alternative geographic comparisons, propensity-score matching, and entropy balancing, with estimates ranging from 43 to 135 percent of the preferred estimate.

Several additional findings support the interpretation that the decline resulted from the ESA’s legal requirements rather than unrelated changes in local housing demand or environmental politics. Restrictions associated with critical-habitat designations produced the largest negative estimates. Plant listings, which generally imposed weaker restrictions on private land, had effects close to zero. Proposed listings also had approximately zero effect before becoming legally enforceable. When species were removed from the endangered species list, permitting increased by roughly the same magnitude that it declined after listings, although estimates based on the relatively small number of delistings are imprecise.

The effects varied across species and local housing markets. Bird listings reduced permitting more than mammal listings, while listings involving clams and some aquatic species produced little to no decline. Listings covering the largest geographic areas reduced annual permitting by approximately 0.98 units per 1,000 residents, whereas species with the smallest habitats had nearly zero effect. The reductions were also larger in places with higher vacancy rates, where expected returns from construction may have been lower and relatively modest regulatory costs may have caused marginal projects to become unprofitable.

Satellite land-cover data show that listings affected both outward development and construction within already-developed areas. Following a listing, greenfield land was approximately 10 percent less likely to be converted into developed land. However, developed open space was also about 9–10 percent less likely to transition to more intensive use. Permits for units in buildings containing five or more residences fell by approximately 16 percent, and permitting reductions were at least as large in highly developed places as in the full sample. These patterns suggest that the ESA did not simply redirect construction away from sensitive undeveloped land and toward infill development.

This distinction matters because restricting greenfield development may directly protect an endangered species’ habitat, while restricting denser infill construction may offer much smaller ecological benefits. Infill can replace existing structures or intensify land that is already developed and unlikely to support endangered species. It may even reduce environmental damage by limiting suburban expansion, transportation use, and energy consumption. The evidence therefore suggests that some ESA restrictions may be poorly targeted, constraining housing in dense cities without producing corresponding improvements in species protection.

To estimate the national consequences, my analysis accounts for the possibility that construction blocked in one jurisdiction was relocated elsewhere. State- and regional-level estimates suggest that approximately 70 percent of the local permit decline represents housing that was deterred altogether rather than displaced. Under the preferred assumptions about displacement and the diminishing effects of additional overlapping listings, ESA regulations displaced or deterred an estimated 9.1 million permits between 1980 and 2024. The central estimate implies that the 2025 national housing stock is approximately 6.3 million units smaller—equivalent to about 4 percent of that stock. Depending on the estimated treatment effect and modeling assumptions, the implied reduction ranges from 0.6 million to 13.8 million units.

Several limitations must be considered. Species listings may still be associated with unobserved local conditions despite the parallel prelisting trends, geographic controls, and matching procedures. The satellite estimates are based on relatively small annual land-use changes and are therefore imprecise. Evidence on large apartment buildings cannot conclusively distinguish infill projects from dense construction on previously undeveloped land. Most importantly, the paper estimates the housing costs of ESA protections but does not measure their benefits for endangered species. The national housing estimates also depend on uncertain assumptions about displacement and how the costs of multiple listings accumulate.

Ultimately, the findings suggest that environmental protection and housing construction may not always require a direct trade-off. The most urbanized 15 percent of permit-issuing places account for approximately 90 percent of housing permits, while the most valuable endangered species habitats generally lie outside these dense areas. Narrowing or simplifying ESA requirements for construction on already-developed land could increase housing supply without substantially weakening habitat protections. By allowing more infill and concentrating development in dense, energy-efficient communities, better-targeted regulation could advance both housing and environmental goals.

Note
This research brief is based on Maxwell Tabarrok, “The Costs of Environmental Land Use Regulation,” Journal of Public Economics 262 (forthcoming).