Roc Armenter, Miklós Koren and Dávid Krisztián Nagy

July 2014


We build a continuous-space theory of trade in which people in a region agglomerate to exploit trading opportunities with another region. The regions are separated by a river, which can be crossed anywhere, but more cheaply at bridges. In the model, most trade takes place via bridges, leading to a key prediction that population density declines with distance to the bridge. We derive additional predictions about the spatial distribution of population and test them on current high-resolution population density data around twelve major American rivers. The data is mostly consistent with our model. In a historical event study of 19th-century bridges on these rivers, we find that the neighborhood of bridges developed faster after the bridge was built. Also, the two sides of the bridge converged in development, highlighting the connecting role of the bridge. More generally, our results suggest that economies of density arising from transport infrastructure can help explain why and where people agglomerate.

Work in progress.


  1. Working paper
  2. Quadtree Python package to efficiently search points within a polygon
  3. Slides