RESEARCH IN PROGRESS

The Effect of Interstates on Regional Development: Novel RD and IV Approaches [slides] 

To estimate the long-run effects of transport infrastructure on employment growth, I develop a novel identification strategy. From planning documents, I discover that cities with populations over 50,000 in 1940 are more likely to be included in the Interstate system. I use this discontinuity to create a bias-adjusted instrument for counties’ distance to an Interstate, following Borusyak & Hull (2023). For many counties, their distance to an Interstate is affected by the plausibly exogenous inclusion of counties near the population discontinuity, however exposure to this exogenous shock is non-random. To adjust for non-randomness, I simulate plausible counterfactual networks by permuting which counties near the cutoff are included in the network and constructing their proximity graph. By controlling for the average distance of counties to these counterfactual networks, I isolate identifying variation that arises from the discontinuity at an arbitrary cutoff.

Highways and Segregation: Evidence from the US South

Highways can have disparate impacts across racial groups, impacting patterns of racial segregation. However, the empirical relationship between dissimilarity indices and the presence of Interstate highways is sensitive to the size of the spatial unit at which race counts are aggregated. Dissimilarity indices constructed from Census tracts yield systematically different results to those constructed from Census block groups. In this project, I examine this further to determine how much of this difference is explained by (1) the scope and scale of racial sorting and (2) bias, arising from endogenous Modifiable Areal Unit Problems.