Working Paper: CEPR ID: DP15470
Authors: Gabriel Ahlfeldt; Thilo Albers; Kristian Behrens
Abstract: Using establishment-level data for the three largest US metropolitan areas, big data for 125 global cities world-wide, and a methodology combining both, we delineate the densest clusters of economic activity. We show that---within cities---these prime locations concentrate 35\% of tradable services employment on 0.3\% of developable land. Although only 40\% of our sampled cities are monocentric, prime locations are---consistent with the theoretical workhorse urban models---the nuclei of distance gradients, even in polycentric cities. Cities with fewer prime locations further concentrate a larger tradable services share in them, underscoring the importance of agglomeration economies for those locations.
Keywords: prime services; internal city structure; transport networks; multiple equilibria; path dependence
JEL Codes: R38; R52; R58
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Major disasters over the 20th century (H84) | More dispersed distribution of prime services (D39) |
| Doubling the population in 1900 (J11) | Decrease in average distance between prime locations (R12) |
| Early development of mass transit systems (L91) | Higher concentration of prime services (G29) |
| Major disasters (H84) | Shift in spatial distribution of prime services (R12) |
| External returns to scale (D24) | Disasters shift spatial distribution of prime services (H84) |