Abstract

A variety of approaches to delineating metropolitan areas have been developed. Systematic comparisons of these approaches in terms of the metro area landscape that they generate are however few. Our paper aims to fill this gap. We focus on Indonesia and make use of data on commuting flows, spatially fine-grained population, and remotely sensed nighttime lights to construct metropolitan areas using several approaches that have been developed in the literature. We find that the maps and characteristics of Indonesia's metro area landscape generated when using a commuting flow approach differ substantially from those generated using other approaches. Moreover, combining infomation on the metro areas generated by the different approaches with detailed micro-data from Indonesia's national labor force survey, we show that the estimated agglomeration wage premium for Java-Bali tends to fall when using a more restrictive definition of metro areas. This is not the case for the rest of Indonesia, for which we, moreover, find a much lower estimated agglomeration wage premium. We provide an explanation for these findings, and also tentatively probe the factors behind Indonesia's agglomeration wage premium.

Highlights

  • Except for the United States, for which data on metropolitan statistical areas (MSAs) are readily available, urban economists have relied on data for cities as defined by their administrative boundaries

  • These three approaches are the Agglomeration Index (AI), which was originally developed by Uchida and Nelson (2009) for the World Bank’s 2009 World Development Report on “Reshaping Economic Geography” (World Bank, 2008); a “Cluster Algorithm” developed by Dijkstra and Poelman (2014) which associates cities with dense clusters of population; and, the identification and delineation of metro areas based on the “thresholding of Night-Time Lights data” (NTL), similar to, for example, Ellis and Roberts (2016) and CAF (2017)

  • One issue that we face in generating results that can be compared across the different approaches for delineating metro areas is that while Duranton’s algorithm uses sub-national administrative units – in our case, Indonesian districts – as the “building blocks” for metro areas, the remaining approaches rely on much higher resolution gridded data sets

Read more

Summary

Introduction

Except for the United States, for which data on metropolitan statistical areas (MSAs) are readily available, urban economists have relied on data for cities as defined by their administrative boundaries. The other three approaches are “second best” approaches which instead rely on global data sets derived, wholly or in part, from satellite imagery These three approaches are the Agglomeration Index (AI), which was originally developed by Uchida and Nelson (2009) for the World Bank’s 2009 World Development Report on “Reshaping Economic Geography” (World Bank, 2008); a “Cluster Algorithm” developed by Dijkstra and Poelman (2014) which associates cities with dense clusters of population; and, the identification and delineation of metro areas based on the “thresholding of Night-Time Lights data” (NTL), similar to, for example, Ellis and Roberts (2016) and CAF (2017).

Approaches to defining metro areas
Data sources
Mapping to districts
Key descriptives of identified metro areas
Jaccard indices and a closer look at levels of agreement
Estimation strategy
Main Sample
Baseline results
Including all districts in the sample
Including only metro districts
Accounting for potential endogeneity of our main city size variables
Findings
Conclusion

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.