This paper studies the long-term impacts of a large temporary demand shock on regional industrialization. I examine the Quartermaster sites through which the Union Army procured general manufactured goods during the U.S. Civil War, placed for wartime logistical reasons and dismantled when the war ended. Procurement had large and persistent effects on the host counties. Relative to comparison counties, by 1890 the QM counties had 66 percent more manufacturing employment and a population 19 percent larger. The manufacturing lead persisted until the 1960s and then declined with the national contraction of manufacturing, while the population lead persisted. The expansion came through the persistent adoption of new industries, and counties became less farm-dependent earlier. A quantitative spatial model in which procurement raised a county’s rate of adopting new techniques rationalizes the empirical findings and implies that procurement raised Union real income per worker by 5.9 percent by 1940.

How do natural advantages shape regional industrial growth? We study the 1892 opening of the Mesabi Iron Range, which within two decades supplied most of the nation’s iron ore, as a plausibly exogenous shock. Using novel data on iron and steel establishments, we compare counties by their transport access to Mesabi ore before and after the opening. We find that by 1926, a one-standard-deviation increase in access had raised a county’s annual blast furnace capacity by about 32,000 gross tons and increased iron and steel employment by about 330 percent. Furnaces at inland ore deposits were largely abandoned, and capacity became concentrated at Great Lakes ports and nearby coal districts. Better access also had large effects on employment in downstream industries like machinery and transportation equipment, along with an increase in total county employment. Resource abundance combined with cheap lake transport substantially shaped the geography and growth of American industry.

During World War II, the United States seized U.S. patents held by residents of the Axis and Axis-occupied countries. Beginning in 1942, the Alien Property Custodian seized about 38,000 patents and licensed them to any American applicant. We digitize the seized patents and construct a measure of each patent subclass’s exposure to the seizure. We find that in exposed subclasses, U.S.-origin patenting rose sharply in 1942, to about 27 percent above unexposed subclasses, and remained 38 percent higher as late as 1975. The increase is larger and more persistent for patents assigned to U.S. firms, and it is not explained by wartime research funding. Patenting in the United States by inventors from the Axis and occupied countries fell by about 78 percent in the exposed subclasses and had not recovered by 1975. The seizure persistently shifted invention in the exposed technologies from foreign to American inventors.

I study industrial policy in its most direct form, in which the state builds and operates manufacturing plants, and evaluate its effects on the regions that received them. The setting is India’s post-independence industrialization drive, in which the primary tool was government-built and government-run manufacturing plants, sited with a stated preference for industrially backward regions. To study this program, I build two new datasets: a boundary-consistent sub-district panel of India from 1941 to 2011, digitized from the Primary Census Abstracts of 1951 to 1981, and hand-collected data on the universe of plants built by the central government. In a staggered difference-in-differences design, I compare sub-districts that received a plant with sub-districts in the same state that never did. Two decades after construction began, population in plant sub-districts is about 20 percent higher and non-farm employment about 50 percent higher. The effects are largest for the earliest cohort of plants.