SAN FRANCISCO — Report after report explain that more than half of the world’s population now lives in urban areas. That number is based on the United Nations World Urbanization Prospects, the most authoritative and cited source of global urbanization levels. But there is debate over what defines an urban area — intensifying in recent weeks amid resurfaced data challenging the U.N.’s calculations — and Devex has learned that the U.N. Statistics Division will convene an expert group this fall to try to reach a common definition.
The U.N. bases its demographic numbers on statistics reported by member states. Its 2018 projections estimate that 55 percent of the world’s population lives in cities, a number it predicts will grow to 70 percent urban by 2050. But researchers at the European Commission are among those who say these widely accepted numbers are incorrect. By leveraging high-resolution satellite imagery, a population grid that divided territory into 1 kilometer by 1 km squares, and a methodology they call the degree of urbanization, they estimated that 84 percent of the world is urban, translating to billions more people living in cities than the U.N. estimate.
See more on urban development:
This is not the first effort to leverage remote sensing or population grids to track urban transformation. But the research is causing a stir, particularly given a recent Thomson Reuters Foundation story headlined: “Everything we’ve heard about global urbanization turns out to be wrong.” Even some of those who disagree with the accuracy of these projections say they are still an improvement over U.N. estimates, and they tell Devex this debate over definitions is an important one to have.
Experts say a universal definition of what is urban may or may not be useful on the national level, where a wide range of definitions are used, but that it would benefit policymakers on the global level working on agendas such as the Sustainable Development Goals — particularly in least developed countries where most of the increase in urbanization is taking place.
When the numbers don’t add up
Every country defines urban differently and collects data according to its own definition. The minimum population threshold of an area classified as urban ranges from 200-50,000 people, according to U.N. World Urbanization Prospects.
“As a result, what is classified as urban in one country may become rural in another,” reads the The State of European Cities 2016, which was released at the 2016 U.N. Conference on Housing and Sustainable Urban Development, or Habitat III, in Quito, Ecuador. “It should not come as a surprise that using a single definition based on population grids yields a sometimes radically different picture.”
Population numbers are critical for everything from planning and delivering services to responding to disasters. The challenge in distinguishing urban versus rural extends beyond the differences in national definitions — it is often difficult to get reliable demographic data in resource-poor settings. Researchers, including the team at the European Commission, are looking to recent advances in the availability of satellite imagery, along with other technological advances, to gain what they believe is a more accurate understanding of populations at both the national and global level.
To apply this degree of urbanization methodology globally, European Commission researchers combine a global population grid, using satellite imagery to detect buildings, with a layer made up of municipal boundaries. They divide populations into three groups — cities, towns and suburbs, and rural areas — and classify cities and towns and suburbs as urban areas.
Combining satellite analysis with census data, the analysts categorize any contiguous stretch with at least 5,000 people and a density of 300 people per square kilometer as urban, supplementing their satellite analysis with census data.
By using this methodology and definition, the research concludes Asia and Africa are respectively 89 percent and 81 percent urban, roughly double the U.N. estimates of 50 and 40 percent.
Yet a recent paper by the Marron Institute of Urban Management at New York University challenges the findings, saying that the European Commission’s numbers come out far higher than the U.N.’s because of the low-density threshold chosen to distinguish an urban from a rural settlement. According to them, at this threshold, a residential plot on the edge of an urban area with a single home sitting on an area as large as two football fields would still be considered urban.
European Commission analysts also acknowledge the limits of their approach, explaining that the current iteration of the grid is most reliable for large cities and developed countries. Lewis Dijkstra, the lead researcher, said that in countries where population estimates are of poor quality, or it is difficult to detect buildings using remote sensing, the margin of error is likely high and these estimates could be off the mark.
“Countries have the right to define a city — we’re not challenging that. What we’re saying is it’s useful to know what this area called urban in one country would be called in another country.”— Lewis Dijkstra, deputy head of economic analysis at the EC’s Regional and Urban Policy division
Still, Dijkstra said that while the grid may be imperfect, the international community has to have alternatives to a global figure generated from data reported by individual countries. Because countries have different definitions of what is urban and do not measure consistently, such a figure cannot be accurate, he said.
“These national definitions are fine within their national contexts, but you can’t add them up. Countries have the right to define a city — we’re not challenging that. What we’re saying is it’s useful to know what this area called urban in one country would be called in another country,” Dijkstra said. “The data we’ve been relying on for so long is basically too distorted to be used for comparisons.”
He added that decisions including the allocation of foreign aid are being made based on urbanization data that is way off. Dijkstra gave the example of international efforts to slow down urbanization in Latin America, because of concerns that the region is becoming too urban for its development, which he called bad advice based on flawed data. And while he acknowledged there is value to countries maintaining their own definitions of urban for their own purposes, he explained that it also useful for them to know where they stand globally.
By 2020, the European Commission researchers hope to get their definition of urbanization recommended for international comparisons. The expert group meeting is a key first step in realizing that goal. But no matter the outcome, Dijkstra said he hopes the research raises awareness of the pitfalls of adding up national definitions, and leads to continued discussion on the value of a universal definition. He also said he hopes these findings will lead more countries to use geocoded data, assigning spatial reference data such as latitude and longitude to locations in upcoming censuses, and building and population registers.
An invitation ‘to do better’
The Marron Institute of Urban Management’s paper called “Our Not-So-Urban World,” published last Wednesday, outlines four arguments rejecting the estimate that 84 percent of the population is urban.
It features a map of the island of Java, Indonesia, and explains how applying the density threshold of 300 people per square kilometer to the map classifies an island of dense villages as if it were a series of large urban areas.
The researchers point to the recent “Everything we’ve heard about global urbanization turns out to be wrong” story, and express concerns that this 84 percent figure will be “repeated ad nauseum by media outlets looking for scoops or by interested parties anxious to make use of this number to further their political agendas.”
What they take issue with is not so much the 5,000 persons threshold, but the urban threshold density of 300 people per square kilometer, explaining that it results in identifying areas such as agricultural regions with small plots tended by farmers, that are clearly not urban, as urban.
“The European Commission is to be applauded for taking the first stab at a method that, in contrast with the U.N. Population Division, uses a common urban population threshold for all countries,” they write. “This should be an invitation to other interested parties to do better and to do better soon. For the time being, however, we recommend that the 84 percent number proposed by the European Commission not be bandied about but be laid to rest, and we urge its competent researchers to go back to the drawing board and start afresh. The work they are engaged in is both valuable and important, and they have both the human and the financial resources to get it right.”
The NYU researchers propose another definition of what is urban, which does not take population density into account, but rather considers the contiguity of built-up areas, and they define urbanization “as the movement of people from being closer to the land to being closer to each other.”
While the U.N. may be way off, said Anjali Mahendra, director of research at the World Resource Institute Ross Center for Sustainable Cities, remote sensing also has its limits. She cautioned that satellite image-based estimates can miss informal settlements, citing research showing that there is variability in interpreting satellite imagery, which can lead to gaps in identifying the populations most in need. Mahendra advocated supplementing these efforts with quick methods of ground truthing, like drones.
“We need an approach — perhaps standardized by the U.N. — that every country can follow to do this on its own once everyone agrees on a standard definition of ‘urban,’” she said.
Dijkstra told Devex their resolution is high enough, and the bar for the number and density of people low enough, that their definition of urban includes informal settlements. He said the only issue with estimating the population of informal settlements is that the European Commission’s methodology also relies on census counts, since satellites can’t count people, and in some informal settlements those are not accurate.
Toward a global definition?
The U.N. Population Division standardizes information across countries in order to come up with internationally comparable population data, such as size and distribution by age as well as changes over time, but while there are some definitional issues when it comes to population data, including marriage or birth or migration, urbanization presents a unique challenge, said John Wilmoth, director of the U.N. Population Division.
“We believe in international harmonization of data and statistics, and that’s really the job of the Population Division in particular to do standardization of international data — so the fact that we haven’t done it in the case of urbanization is a reflection of how difficult it is,” he said. “What I want is a robust discussion of this topic to see if it would be possible to arrive at a globally agreed definition, and it’s not completely clear to me that it’s possible — but if it’s possible, I think it would be valuable.”
Because the U.N. is bound by its mandate to use only data provided to it by countries, and every country has its own definition of what is urban, aggregating the data is adding apples and oranges, said Solly Angel, the lead author of the NYU paper.
While he does not agree with the European Commission methodology, he said he does hope the international community will come to agreement on a definition, which will allow leaders to make accurate comparisons and better define policies.
Angel also said he is in favor of the expert group meeting discussing a universal definition of urbanization, but as a first step, proposed making national methods of deciding what constitutes urban more transparent. If the international community understood the reasoning behind the different definitions, they could still model countries into a common threshold — even if national definitions continue to vary widely, he said.
“It is very difficult to believe that we are going to come to one size fits all.”— Srdjan Mrkić, chief of demographic statistics at the U.N. Statistics Division
On a phone call with Devex, Srdjan Mrkić, chief of demographic statistics at the U.N. Statistics Division, opened up a UN Demographic Yearbook. Every year since 1948, the U.N. has collected data from national statistics offices, and the attempts to define what is urban have gone on for almost as long. Mrkić read through a range of definitions of what is urban, which varied widely from Argentina to Azerbaijan, and everywhere in between, then talked more about the upcoming expert group meeting, where he said 18-22 countries will be represented.
“This is the third time that I am personally trying to put together an international understanding of what is urban and rural. The first two times, it failed. There was simply no possibility of reconciling the differences between approaches,” he said. “It is very difficult to believe that we are going to come to one size fits all.”
Beyond the European Commission and NYU, other past attempts to craft a standard urban definition include a methodology from the Organization for Economic Co-Operation and Development, which takes into account factors such as what percent of workers commute to the core.
If the Statistical Commission, which is made up of chief statisticians from 24 member states and meets in March, endorses a standard definition of urbanization, the U.N. Statistics Division would translate the endorsement into a U.N. recommendation, and ask member states to adopt it and report accordingly, Mrkić explained.
Mrkić said he is familiar with both the European Commission and NYU approaches to urbanization, and he said he sees value in both sophisticated modeling techniques, even though the European Commission definition is the one that has generated headlines recently.
“To say one is better than the other is going back to square one: What is the definition of urban?” he said.