Evidence for localization and urbanization economies in urban scaling

Somwrita Sarkar, Elsa Arcaute, Erez Hatna, Tooran Alizadeh, Glen Searle, Michael Batty

Research output: Contribution to journalArticlepeer-review


We study the scaling of (i) numbers of workers and aggregate incomes by occupational categories against city size, and (ii) total incomes against numbers of workers in different occupations, across the functional metropolitan areas of Australia and the USA. The number of workers and aggregate incomes in specific high-income knowledge economy-related occupations and industries show increasing returns to scale by city size, showing that localization economies within particular industries account for superlinear effects. However, when total urban area incomes and/or gross domestic products are regressed using a generalized Cobb-Douglas function against the number of workers in different occupations as labour inputs, constant returns to scale in productivity against city size are observed. This implies that the urbanization economies at the whole city level show linear scaling or constant returns to scale. Furthermore, industrial and occupational organizations, not population size, largely explain the observed productivity variable. The results show that some very specific industries and occupations contribute to the observed overall superlinearity. The findings suggest that it is not just size but also that it is the diversity of specific intra-city organization of economic and social activity and physical infrastructure that should be used to understand urban scaling behaviours.

Original languageEnglish (US)
Article numberrsos191638
JournalRoyal Society Open Science
Issue number3
StatePublished - Mar 1 2020


  • Agglomeration economies
  • Occupational concentrations
  • Urban scaling
  • Urbanization and localization economies

ASJC Scopus subject areas

  • General


Dive into the research topics of 'Evidence for localization and urbanization economies in urban scaling'. Together they form a unique fingerprint.

Cite this