Abstract: Flooding is one of the most pervasive and damaging natural disasters in the world, and its severity and frequency is expected to increase with a changing climate. Understanding how that increased risk can be expected to affect the economic behavior of flood-exposed regions is therefore of particular importance. In this paper, I look to the effects of flood control dam construction in twentieth-century America to gain insight into how economic behavior may adjust to a changing natural disaster landscape. During most of the twentieth century, the US constructed hundreds of large dams across the country, thereby lessening downstream flood risk by allowing for a greater degree of waterway management. To examine how counties responded to that changing flood risk profile, I first construct a novel dataset that measures the degree to which dams provided flood control for the counties downstream from them. Using an identification strategy that exploits the fact that dams control large extents of rivers, making it so that there were many locations downstream of dams that were not their intended beneficiaries but nonetheless experienced a flood risk reduction, I find that flood control led to an increase in per-acre agricultural land and equipment values. My estimates imply that, aggregated over all counties in my sample, flood control was responsible for $20 billion of agricultural land value in 1992 (2022 dollars), or 5% of the total in-sample land value in that year. Flood control accounted for $1.7 billion (2022 dollars) in aggregate equipment value in 1992, or 2.7% of the total. These results point to the possibility that places facing heightened disaster risk in the future, similar to pre-flood control America, may experience a dampening of productivity expectations and investment. At the same time, the US flood control experience highlights the possibility and importance of natural disaster adaptation technologies.

Abstract: Using weekly mortality data for London spanning 1866-1965, we analyze the changing relationship between temperature and mortality as the city developed. Our results show that both warm and cold weeks were associated with elevated mortality in the late 19th-century, but heat effects, due mainly to infant deaths from digestive diseases, largely disappeared after WWI. The resulting change in the temperature-mortality relationship meant that thousands of heat-related deaths–equal to 0.8-1 percent of all deaths–were averted. Our findings also indicate that a series of hot years in the 1890s substantially changed the timing of the infant mortality decline in London.