Subways and Urban Growth: Evidence from Earth

We investigate the relationship between the extent of a city's subway network, its population and its spatial configuration. To accomplish this investigation, for the 632 largest cities in the world, we construct panel data describing the extent of each of the 138 subway systems in these cities, their population, and measures of centralization calculated from lights at night data. These data indicate that large cities are more likely to have subways, but that subways have an economically insignificant effect on urban population growth. Consistent with economic theory and with other studies of the effects of transportation improvements on cities, our data also indicate that subways cause cities to be more decentralized. For a subset of subway cities we also observe panel data describing subway and bus ridership. We find that a 10% increase in subway extent causes about a 6% increase in subway ridership and has no effect on bus ridership. Consistent with the available literature describing the effect of roads on cities, our results are consistent with subways having a larger effect on the configuration of cities than on their sizes, and with subways having a larger effect on discretionary than commute travel.