1. Developing a global, people-based definition of cities and settlements
By Lewis Dijkstra,
Head of the Economic Analysis Sector
DG for Regional and Urban Policy, 1

2. SDG City goals, but no city definition
City Centre
Edge of the city
Open Space
Low
High
Air pollution
Access to transport
Built-up area per head
Population change
Low (neg)
Built-up change

4. The well known narrative…
The world is 50% urban.
Middle income countries are less urbanised.
Low income countries are least urbanised.
Urbanisation will grow rapidly in low-income countries between 2010 and 2050.
… is purely based on national definitions

5. But national definitions are fine
I know it when I see it (USA Sup. Court)
Differences between national definitions must be quite small.
A city is a relative concept and differs between cultures and countries.
But do we see the same things?
Can this be verified without a global definition?
Do economies of transport, scale and agglomeration require more people in some countries?

6. A single density threshold cannot reproduce the nationally defined urban population shares

7. National definitions vary and are often not statistical
75 countries use population size or density, but thresholds and spatial units vary
47 use a combination of population and other indicators
10 use other indicators than population
100 countries use administrative designations, not a statistical definition that can be replicated in other countries

8. Around half the national definitions rely on population size (and density)

9. 85 countries out of 103 use a minimum population of 5,000 or less

10. 14 countries use density
Six other countries did not report the density threshold
Density is highly dependent on the size of the spatial unit

11. Size of grid cell in sq km

12. What indicators can be used globally
Agricultural employment? Varies too much
Infrastructure? No harmonised data
Services? No harmonised data
Poverty? Circular argument
Rural cannot be defined by problems, because then no problems = no rural
Remoteness Separate dimension

13. Access to cities - Remoteness

14. Two definitions joined by cities
Functional Urban Area
Cities
Towns & suburbs
Rural areas
Commuting zones
Non-metro areas
Degree of urbanisation

15. Two different concepts of urban

16. Urban areas … lost in translation?
Small urban areas
Large urban areas
Medium density
Medium population
Medium share of agricultural jobs
Some services (primary school, doctor)
Europe & Americas
High Density
Big population
Low shares of agricultural jobs
Specialised services (higher education, hospital, government)
Africa & Asia

17. National definition more urban (green) Degree of urbanisation more urban (red)

18. More agreement on large cities

19. 77% (98%) of nationally defined cities match an urban centre of 300,000+ (all urban centres)

20. 90% (99%) of the population of nationally defined cities matches an urban centre of 300,000+ (all UCs)

21. 580 (35%) urban centres of 300k+ do not match a nationally defined city

22. 12% of the population of urban centres of 300k+ is not in a city

23. Pilot projects to compare definitions
Apply the definitions also to administrative units (census tracts, municipalities…)
Appraise the result:
Too urban or too rural?
Too many or too few cities?
Improve data (better population and/or remote sensing data)
Find out if/how we can improve the definitions

24. Pilot projects by EC, OECD and WB
Australia
Brazil (completed)
Colombia
Egypt
Haiti
Indonesia
India
Jordan
Malaysia
Mozambique
Pakistan
South Africa (completed)
Tunisia
Turkey
Uganda
Ukraine
USA

25. Australian cities

26. Colombian cities

27. Brazil
We believe that this method offers a useful basis for statistical comparisons across national borders...
… useful for generating Sustainable Development Goals’ indicators, producing data for these three types of cluster or for individual municipalities …

28. Survey by UN Statistical Division
Algeria
Argentina
Australia
Bolivia
China
Cuba
Ecuador
Ethiopia
Indonesia
Japan
Mexico
Mongolia
Namibia
New Zealand
Republic of Korea
Senegal
Thailand
USA
Venezuela
Zambia

29. Responses to the UN survey
9 out of 12 NSIs: it captured their main cities.
5 out of 8 NSIs the validity was good or satisfactory (1 poor and 2 unacceptable)
9 out of 10 NSIs could produce data by degree of urbanisation
9 out of 13 NSIs useful for international comparisons
11 out of 12 NSIs useful for measuring the SDGs
5 NSIs did not reply
Some confusion about density thresholds and the distinction between cities and towns

30. Driving time buffers Commuting data is not widely available
We will create a driving time buffer around each urban centre. Distance will depend on the population size of the city, GDP per head of the country and the road network
Time series in selected countries

31. Timeline Launched at the Habitat III conference in 2016
Side event UN Statistical Commission
2017 as part of the UN GGIM meeting
2018 an event dedicated to this topic
Plenary Meeting of the UN Statistical Commission
2019: for information
2020: for discussion and decision

32. More information