Conference Agenda

I examine the long-run consequences of a temporary, localized energy boom on local manufacturing development. Exploiting the 1886 discovery of natural gas in eastern Indiana, I use decadal U.S. census data on manufacturing activity to document how fifteen years of cheap natural gas spurred manufacturing growth in the region that persisted for nearly a century—despite the gas supply’s collapse in 1901. Comparing gas-rich counties with similar non-gas counties, I show that gas counties experienced both immediate and sustained manufacturing growth. Initially driven by glass and iron industries, this growth survived the collapse of gas supply, and later expanded to machinery and automobiles, suggesting the presence of agglomeration forces in manufacturing that offset the loss of the initial advantage. This early industrial boom in the gas region was driven by a unique geological circumstance that complicated the transportation of gas to distant markets and pushed gas producers to attract manufacturers to the region with offers of cheap energy. To underscore the importance of this transportability constraint, I contrast Indiana with the Appalachian gas region, which faced no such limitations and exported most of its gas to distant urban markets. Appalachian gas counties experienced only modest excess manufacturing growth. The comparison highlights that even a temporary resource boom, when used as an input in local production, can create enduring advantages for local manufacturing industries.

Rising regional inequality has renewed interest in place-based infrastructure investments as a tool for promoting local economic development. This study uses the large-scale roll-out of electric transmission infrastructure in Nigeria from 2009 to 2015 to quantify the effect of electrification on internal migration. We address endogenous allocation of electricity infrastructure by estimating effects on peripheral households not directly targeted by the policy, in combination with instrumenting for the actual grid path using a hypothetical least-cost grid. Results show an increase in individual migration propensity by 6 percent and a reduction in household size by 0.8 individuals, mainly driven by young adults and older teenagers. These effects are consistent with electrification relaxing credit constraints while generating limited local employment opportunities for younger individuals. Results from a gravity model of migration show a reduction in the elasticity of migration with respect to movement costs and a rise in migration to rural, electrified destinations following the electricity supply shock. Taken together, these findings suggest that productivity-enhancing place-based infrastructure investments may facilitate out-migration from disadvantaged areas and have unintended implications for regional inequality and urbanization.

The lack of an emerging consensus in the large body of research examining how home prices respond to natural disasters presents a puzzle: Why do some papers find increases while others find decreases? We use property-level disaster risk, treatment, and real estate listing and sales data to examine the dynamics of housing market supply and transactions after hurricanes. We find that after a temporary initial disruption in listing and sales activity, hurricanes cause composition shifts of the types of homes being supplied to the market and of the types of homes that clear the market, potentially introducing ``transaction bias,'' a form of sample selection bias. Starting several months after hurricane landfall, listings for flooded homes increase but sales do not. Flooded homes listed after landfall are more likely to have elevated foundations and less likely to have upgraded features. We then perform a novel analysis that eliminates sample selection bias by focusing on listings that were started shortly before hurricane forecasts.

In this sample, flooded homes are nine percent less likely to sell while elevated homes in the wider region are 2 percent more likely to sell. Listings for non-elevated flooded homes that do end up selling spend over a month more on the market and are nine percent more likely to sell at large discounts. These compositional shifts indicate that extreme weather events can cause sample selection bias in sales, a phenomenon known as ``transaction bias'' that can skew price effect estimates and may explain some of the disagreement in the price effect research.

Climate change is expected to reduce agricultural productivity in developing countries. Classic international trade and geography models predict that the optimal adaptation response is a reallocation of capital and labor from agriculture towards sectors and regions gaining comparative advantage. In this paper, we provide evidence on the effects of recent changes in climate in Brazil to understand to what extent factor market frictions constrain this reallocation process. We document that persistent increases in dryness do not generate capital reallocation but a sharp reduction in credit to all sectors in both drying areas and financially integrated regions. In addition, dryness generates a large reduction in agricultural employment. Workers staying in drying regions reallocate towards manufacturing, but climate migrants are allocated to small firms outside of manufacturing in destination regions. The evidence suggests that frictions in the interbank market and spatial labor market frictions constrain the reallocation process from agriculture to manufacturing.