Capital Cities and Road Network Integration: Evidence from the U.S.

This paper quantifies the causal effect of capital status on road network integration of U.S. micro/metropolitan statistical areas. Road network integration is defined using a class of measurements that evaluate how well a location is connected to all other locations through the National Highway System (NHS). To tackle the non-random placement of capital cities, I instrument capital status using a k-means clustering algorithm that predicts the boundaries of 48 U.S. states and defines the geographical center as a hypothetical capital location. Overall, I find significant and robust evidence that capital cities are more directly integrated in the NHS than non-capital cities of similar characteristics. I discuss two possible mechanisms behind the capital premium: (i) the favorable geographical position of capital cities within their state and (ii) a political interest in connecting capital cities well to major urban areas around.