Because national definitions of urban and rural areas differ significantly from one country to another, it is difficult to compare these areas across national borders. If we can’t compare the performance of urban or rural areas across national borders, then we can’t learn from policies used in other countries. It also means we cannot meaningfully compare the United Nations (UN) Sustainable Development Goals' (SDGs) indicators for urban and rural areas across countries.
On March 5th, the UN Statistical Commission endorsed the Degree of Urbanization as a recommended method for international comparisons.
Many countries use a minimum population size to define an urban area, but that size can be 200 (as in Denmark), 2,000 (Argentina), 5,000 (India) or 50,000 (Japan) or even 100,000 (China). Some countries don’t use a statistical definition but designate urban areas by administrative decision. In other countries, the sectoral employment or provision of infrastructure and services is used to determine whether settlements should be classified as urban or rural.
Finally, once categorized as urban or rural, places are rarely recategorized. Some of this resistance may come from the allocation of fiscal transfers – consider India, where getting reclassified as urban may cause places to lose government transfers, or Egypt, where getting reclassified as urban would trigger additional public investment for higher-level service delivery requirements, including police stations and courthouses.
A wide-angle view to measure urbanization
, the Degree of Urbanization seeks to do for the definition of urban what the $1/day poverty line did for poverty measurement in the 1990s.
The Degree of Urbanization identifies three types of settlements:
- Cities, which have a population of at least 50,000 inhabitants in contiguous dense grid cells (>1,500 inhabitants per km2);
- Towns and semi-dense areas, which have a population of at least 5,000 inhabitants in contiguous grid cells with a density of at least 300 inhabitants per km2; and
- Rural areas, which consist mostly of low-density grid cells (<300 inhabitants per km2).
This new approach offers several advantages:
- Simplicity and transparency. It relies on the simple combination of population size and density applied to the population grid, instead of a multitude of criteria or complex and lengthy calculations. An increasing number of countries have their own population grid. Several global population grids have been estimated and are available for free, including the Global Human Settlement Layer Population Grid (GHS-POP). The estimated Degree of Urbanization for each country in the world using GHS-POP can be found here.
- Driven by population size and density. Population size is used by more than half of the national definitions of urban and rural areas. The thresholds used in the Degree of Urbanization take inspiration from these national definitions. However, it uses two thresholds instead of one. For cities, it uses 50,000 inhabitants as Japan does. For towns and semi-dense areas, it uses 5,000. Out of the 100 countries that use population size threshold, 85 use the 5,000 threshold or a lower threshold. The thresholds used in the Degree of Urbanization were also tested to ensure that they produce a valid and robust classification and a balanced population distribution across the three classes.
- It helps monitor progress on the SDGs. The SDGs include a multitude of indicators that should be collected for cities, urban, and rural areas, including access to electricity, water, the Internet, and all-weather roads. Some definitions of urban areas, however, include access to water and electricity. This makes it impossible to monitor these services in urban areas because it becomes a circular argument. All urban areas have water because by definition they can only be urban if they have access to water. For example, the definitions used by Bangladesh, Cuba, and Panama all include access to drinking water. Because the Degree of Urbanization does not include services or infrastructure, it can monitor these services in an unbiased manner.
- It captures agglomeration economies. Because the definition relies on the spatial concentration of the population, it captures the logic of agglomeration economies. The cost of service provision tends to increase from cities to towns and semi-dense areas and then to rural areas. As a result, access to these services tends to be highest in cities and lowest in rural areas. For example, access to the Internet and having a bank account both follow a clear urban gradient in all four country income groups (Figure 1 and Figure 2).
- Cost-effective monitoring. The Degree of Urbanization can be used to re-aggregate existing data. For example, if a statistical office has measured local employment rates, then it can calculate employment rates by Degree of Urbanization. Geo-coded micro data can also be aggregated. This was done for the Gallup World Poll in 115 countries and for the Demographic and Health Survey in 41 countries in a new report by Vernon Henderson et al. This allowed us to show that access to safely managed drinking water is higher in cities, second highest in towns and semi-dense areas, and lowest in rural areas (Figure 3) in almost all of the countries covered by this survey.
Comparing the Degree of Urbanization with other approaches
As with any newly proposed method, it raises questions, and people want to know how it compares to existing methods. The main questions raised by national statistical offices and academics are addressed below.
- Is the estimated rural population too low?
Applying the Degree of Urbanization to the global population grid GHS-POP results in an estimated rural population share in 2015 of 24%, which is considerably lower than the 46% based on national definitions (Figure 4). The main cause for this 
difference is that 12 large countries classify towns as rural areas. China and India account for half of the difference in rural population. China’s definition makes it explicit that towns and small cities are not considered urban, because they use a threshold of 100,000 inhabitants for urban areas. India uses a threshold of 5,000 inhabitants but combines it with other criteria, which leads to most towns being classified as rural. A further 10 countries account for 30% of this difference: Bangladesh, Democratic Republic of the Congo, Egypt, Ethiopia, Indonesia, Nigeria, Pakistan, Sudan, Uganda, and Vietnam. In short, 12 countries account for three-quarters of the difference between the rural population shares.
In some countries, population data with a high spatial resolution was not available. Applying this method to a grid based on more precise data may lead to higher rural population shares.
- Should a definition of urban and rural areas include agriculture employment?
Thirty-seven countries have a national urban area definition, which includes a maximum share of employment in agriculture. These countries are primarily located in Africa and Asia. The employment share in agriculture, however, differs substantially between countries and by level of development. In high-income countries, it is 3% compared to 63% in low-income countries (Figure 5). Including such a criterion in a global definition would lead to entire countries becoming rural or urban, which would be an obstacle to comparisons. Furthermore, this share is dropping quickly. In 2000, 40% of global employment was in agriculture. In 2018, it had dropped to 28%. As a result, including this criterion would also reduce the comparability over time.
- Should a definition of urban and rural areas be based exclusively on built-up areas?
Historically, the data on buildings had a higher spatial resolution than data on people. As a result, definitions used buildings as a proxy for the spatial concentration of population. For example, several Nordic countries define a settlement as buildings less than 200 meters apart. Several academics also use buildings or built-up area – see, for example, the Urban Extent used by Shlomo Angel, Africapolis and the building density method developed by Marie-Pierre de Bellefon et al. 2019. With the improvements of population data, however, this indirect approach is no longer necessary.
Furthermore, the amount of built-up area per capita is closely linked to the income of a country and will distort the population share in cities and rural areas. To show this, we defined cities as cells of 250 by 250 meters that are at least 50% built-up and rural areas as cells that are less than 25% built-up. These thresholds are used in several built-up area definitions of urban and rural areas.
Compared to the Degree of Urbanization, cities defined by built-up area alone have a population share that is 17 percentage points higher in high-income countries and 9 percentage points lower in low-income countries (Figure 6). For rural areas, using built-up area increases the rural population share by 24 percentage points in low-income countries (Figure 7) and reduces it by 2 percentage points in high-income countries. This makes built-up-area-based definitions of urban and rural areas less suitable for international comparisons.
- Should a definition of urban and rural areas rely on relative thresholds, not absolute ones?
Some academics have argued in favor of using relative thresholds to define urban and rural areas. This could mean, for example, identifying the 10 most urbanized areas in a country. However, a global definition should ensure a high level of comparability across both space and time. Comparing the 10 most urbanized areas in one country with the 10 most urbanized in another does not ensure that these areas have the same level of urbanization. The level of urbanization in the globe is increasing, but a relative threshold cannot capture that, either. Therefore, relative thresholds do not allow for good comparability across space and time.
By standardizing the classification approach and applying it at the global level, it can help identify and measure the effectiveness of the policies in different countries that improve the quality of life in these areas. It will also help monitor access to services and infrastructure and other SDG indicators in a way that enables meaningful comparison and aggregation. This method will be further tested and implemented in many countries around the globe.
We welcome comments and questions regarding the Degree of Urbanization.
- Subscribe to our Sustainable Cities newsletter
- Follow @WBG_Cities on Twitter
Classifying urban areas according to population density is good, but what about industrial areas, don’t they count as urban areas?
The method includes industrial areas in two ways. At the grid level, cells that are 50% built-up can be added to the urban centre. This means that a contiguous industrial area would be included. At the local unit level, industrial areas are usually included in a wider administrative area. If the majority of the residential population of that administrative area lives in an urban centre, it will be classified as a city including the industrial area.
an urban area definition i think depends from place to place eg china where the population is large it can be defined using designated number of people
National definitions already use population size thresholds. The benefit of this method is that it uses a single threshold for cities and a single threshold for towns. In that way, we can compare if all cities in the world have a hospital for example. It would be unfair if we compared settlements with at least 2,000 inhabitants in one country to settlements of with at least 100,000 another.
Hi excellent analysis and insightful comments. If possible, can you draw some attention to the oxymoron of Urban Villages. Where do they come from and what's their future ? Thanks
I welcome with great interest this new approach. Indeed, its simplicity is its strength. Population and density are understandable and usable concepts all over the world. As you clearly explain, its adds transparency and will allow a much more effective international monitoring process.
However, what we gain on the one hand, we lose it on the other hand. We gain easy worldwide monitoring of geographical characteristics (population settlements) and of general global objectives (like SDGs) which is, once again, a great step forward.
But we lose policy relevance.
Concepts of rural, urban or suburban spaces are used by policy makers to tailor social, economic or environmental measures. And population density is hardly ever the good driver for that.
If definitions are so different between countries, it is because they are carrying diverging political vision of space that are meaningful in specific contexts.
And unfortunately, when international comparisons reduce this diversity, they become less useful at national/territorial level.
Functionality is the compass which is guiding territorial policies today and geographical delineations based on settlement are simply not sufficient to drive policy making anymore.
The method is designed to classify administrative areas. These can be municipalities, but also groups of municipalities that work together to provide common services. This matching allows the definition to relate to boundaries used by policy makers. In some cases, these levels will match well. In others, they will not. For example, if an urban centre covers many different municipalities, these municipalities may want to work together to organise public transport and land use. The functional urban area definition is particularly useful for policy making given that it captures an area is likely to cover multiple functions: commuting, education, health care, culture…
Very insightful writeup on Urban areas, lets keep the conversation going. As researchers we need to develop a tool applicable across the board world over.
Thank you so much for the very informative blog and intriguing arguments. I found the suggested methodology simple and appealing, however for its wider adoption, I believe we need to insure its relevance from the perspective of national policy makers’ as well. In Egypt for instance, according to the proposed methodology, most desert settlements that are declared as cities will be reclassified as villages. This would be the correct if we overlooked the role these settlements play in relation to their surroundings settlements who are depending on them for the provision of key urban services- such as secondary schools or general or specialized hospitals. In doing so, including a criteria that consider the “City Region” might be of relevance. In doing so, settlements could be named as cities or towns if they are the sole provider of urban services to their neighboring communities within a radius of 15 or 30 kms, even if they does not qualify for the proposed population and density measures.
Indeed with this method, some cities or towns will not have secondary schools or specialised hospitals and some villages will. We think the presence (or absence) of these services should be used to characterise these settlements, but not change their class: for example "cities without a secondary school", "villages with a general hospital". By not including these services in the definition of what is a city, a town or a village, these services can be monitored more accurately. For example, it allows us to answer the question 'How many towns lack a secondary school?' In addition, the largest settlement within a certain radius could also be described as a local or regional centre. This would show which village or town is likely to provide services to a wider area.
Am I understanding correctly that your urban area definitions are based on long-term residency of persons?
This can be misleading in countries where vast numbers of workers live in urban areas, either only during part or all of the work week, or only for specific work contracts, and then either return to a rural area, or move to another urban center.
This is widely seen in China, but also in some African and South American urban areas.
My research has confirmed this for instance in Western Nigeria and Central Kenya.
Hence urbanization can be a much more hybrid concept than we may hitherto have assumed.
This definition is relies on the census 'de facto population,' not the 'de jure population'. This implies that some people would be counted where they live during the working week and others where they live on the weekend depending on when their census interview takes place. People living in two places are unlikely to have a big impact on this definition. It only takes 50,000 people to become a city. Most places that a have significant number of part time residents would be a city even if all the part time residents were not counted. In countries without good de facto population data and a significant number of people living in a city either part time or without being officially registered, the de jure population will underestimate the total population size of a city. Data from mobile phones could be used to estimate the 'de facto population' where this is missing.
Thanks for explaining urban and rural