Working Paper: CEPR ID: DP13359
Authors: Maarten Bosker; Mark Roberts; Jane Park
Abstract: A variety of approaches to delineate metropolitan areas have been developed. Systematic comparisons of these approaches in terms of the urban landscape that they generate are however few. This paper aims to fill this gap. The paper focuses on Indonesia and makes use of the availability of data on commuting flows, remotely-sensed nighttime lights, and spatially fine-grained population, to construct metropolitan areas using the different approaches that have been developed in the literature. The analysis finds that the maps and characteristics of Indonesia’s urban landscape vary substantially, depending on the approach used. Moreover, combining information on the metro areas generated by the different approaches with detailed micro-data from Indonesia’s national labor force survey, the paper shows that the estimated size of the agglomeration wage premium depends nontrivially on the approach used to define metropolitan areas.
Keywords: metro areas; urban definitions; agglomeration economies; indonesia
JEL Codes: O18; O47; C21
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Definition of metropolitan areas (R11) | Estimated agglomeration wage premium (J39) |
| Commuting flow thresholds (L91) | Estimated agglomeration wage premium (J39) |
| Satellite data (Y10) | Estimated agglomeration wage premium (J39) |
| Incorrect definitions of metropolitan areas (R11) | Bias in wage estimates (J31) |
| Commuting flows and population density (R23) | Causal analysis (C22) |