Abstract
Abstract Bridges are critical but sparse links in land transport networks. I exploit quasi-experimental variation in bridge construction over major rivers in the U.S. to measure the causal effects of land transport infrastructure. Bridges are more often built upstream than downstream of tributary confluences—where smaller rivers join larger rivers—generating local differences in connectivity. Local connectivity advantages have negative effects on per capita income. In contrast, more substantial changes in connectivity arising from the opening of major bridges increase per capita economic activity. A narrative explanation that can reconcile both results is that land transport infrastructure creates productivity advantages that drive economic growth, structural transformation, and urbanization over large spatial scales, but local sorting within the cities that form around early transport routes then reverses this gradient over smaller spatial scales.