Population Centers and Coordination : Evidence from County-Seat Wars

We study the process of long-run urban growth using a unique setting of close elections that determined “county seats” (capitals) in the frontier United States. Employing a regression discontinuity design, we show that winning towns rapidly became the economic and population centers of their counties as new migrants coordinated on them as destinations. This coordination was largest in the early years of a county’s history, but limited in later decades. Using generalized random forests, we show that the economic changes were not zero sum locally: specific choices of county seat could increase long-run county population and income. As county administration was limited in this era, the public sector did not play a substantial role in this growth. Instead, these results illustrate how a political process can select spatial equilibria through a shock that is neither related to locational fundamentals nor confers direct productivity advantages on the location.