Average temperatures for the median U.S. county rose by 0.6°C between the 1980s and the 2010s, and even optimistic mitigation scenarios project a 50% increase in the number of days with maximum temperature above 30°C by century’s end. How a warmer climate reshapes the economy is therefore an increasingly urgent policy question. Drawing on U.S. Census Bureau plant-level data on U.S. manufacturers from 1977 to 2019 combined with high-resolution daily temperature data, we show that the manufacturing economy does not simply shrink under heat — it consolidates. The stock of smaller plants shrinks over time, workers reallocate to larger plants, and local industries become more concentrated. Climate adaptation is thus not only an environmental issue, but also a question of firm dynamism and market structure.
In the short run, year-to-year temperature shocks raise costs and lower productivity, but only for some plants. One standard deviation more days with maximum temperature above 30°C (about 44 additional such days in a year) raises a plant’s energy costs by 4.6% and lowers productivity by 2.7–3.5%. These effects are concentrated in small and medium plants; large plants are largely unaffected. Plant size and closure rates do not respond contemporaneously.
Over the long run, exploiting four decades of variation in average temperature across U.S. commuting zones, a clearer reallocation emerges. A one standard deviation higher increase in cooling degree days — the annual sum of degrees by which mean daily temperatures exceed 18°C, here about 60 additional degree-days per year between the 1980s and 2010s — is associated with a 3.2% decline in the number of small plants and a 6% increase in average plant size. Total commuting-zone employment is essentially unchanged, which means workers are shifting from small to large plants rather than exiting manufacturing. As a result, local industries become more concentrated: the employment share of the top five plants rises by 0.5 percentage points, and the local Herfindahl-Hirschman Index increases by an additional 3%.
Why are large plants better adapters? We find evidence for three mechanisms. First, larger plants pay lower prices per unit of energy, blunting the cost shock from heat; consistent with this, the warming–concentration link is strongest in industry-regions where energy prices are higher. Second, larger plants invest more in adaptation (such as energy-efficient machinery) and adopt formal energy-management practices, and where such practices are widespread, long-run warming has less impact on concentration. Third, larger plants enjoy easier access to finance: using local bank-branch density as a proxy for credit access, we find that better-financed areas show smaller adaptation-driven shifts in concentration, consistent with financing helping smaller plants absorb shocks and invest.
These findings have implications for how policymakers think about climate adaptation. Adaptation policy is typically framed as an environmental problem: reduce emissions, protect against weather damage, and fund resilience. Our findings argue for a broader frame. A warmer climate is quietly tilting local manufacturing toward larger plants and more concentrated industries, and an adaptation policy that focuses only on aggregate employment or output will miss these distributional consequences. Each of the three mechanisms we identify is also policy-actionable: energy-price relief and efficiency subsidies targeted at small and medium plants can offset their disproportionate cost burden; technical assistance can broaden participation in energy-management practices; and improving credit access in less financially developed regions can give smaller plants the room to invest in adaptation. Sustaining the small-plant segment of manufacturing is, in this sense, not only a competition objective but also a piece of climate resilience — one that keeps local labor markets diverse and dynamic as the climate continues to warm. In a warmer economy, who adapts is as important as how much. Adaptation policy should reflect that.