We document a Kuznets curve for construction productivity in 20th-century America. Homes built per construction worker remained stagnant between 1900 and 1940, boomed after World War II, and then plummeted after 1970. The productivity boom from 1940 to 1970 shows that nothing makes technological progress inherently impossible in construction. What stopped it? We present a model in which local land-use controls limit the size of building projects. This constraint reduces the equilibrium size of construction companies, reducing both scale economies and incentives to invest in innovation. Our model shows that, in a competitive industry, such inefficient reductions in firm size and technology investment are a distinctive consequence of restrictive project regulation, while classic regulatory barriers to entry increase firm size. The model is consistent with an extensive series of key facts about the nature of the construction sector. The post-1970 productivity decline coincides with increases in our best proxies for land-use regulation. The size of development projects is small today and has declined over time. The size of construction firms is also quite small, especially relative to other goods-producing firms, and smaller builders are less productive. Areas with stricter land use regulation have particularly small and unproductive construction establishments. Patenting activity in construction stagnated and diverged from other sectors. A back-of-the-envelope calculation indicates that, if half of the observed link between establishment size and productivity is causal, America’s residential construction firms would be approximately 60% more productive if their size distribution matched that of manufacturing.
