Within US Trade and the Long Shadow of the American Secession

Using data from the US commodity flow surveys, we show that the historical Union-Confederacy border lowers contemporaneous trade between US states by about 16 percentrelative to trade flows within the former alliances. Amongst one million placebos, thereis no other constellation of state grouping that would yield a larger border effect. Thefinding is robust over different econometric models, treatment of the rest of the world,available survey waves, or levels of aggregation. Including contemporaneous controls,such as network, institutional or demographic variables, and Heckscher-Ohlin or Linderterms, lowers the estimate only slightly. Historical variables, such as the incidence ofslavery, do not explain the effect away. Adding US states unaffected by the Civil War,we argue that the friction is not merely reflecting unmeasured North-South differences.Finally, the estimated border effect is larger for differentiated than for homogeneousgoods, stressing the potential role for cultural factors and trust.