What makes a place urban?
Abstract
We argue that “urbanity” is a function of population concentration. Empirically, this means
urbanity can be measured along a spectrum, with increased population concentration in a
place denoting increased urbanity. Phenomenologically, large concentrations of people in
space—or mass corporeal co-presence—generates the essentially “urban” experience of living
surrounded by the bodies and minds of strangers. We use a series of thought experiments to
demonstrate the conceptual limitations of other historically common definitions of urbanity,
such as the economic structure of a community, the presence of physical infrastructure, the
political or administrative status of a geographic unit, or the degree of connectivity between
people. These are not essential urban characteristics, but rather common epiphenomena
associated with places that have large, spatially concentrated populations. A density-based
definition does not require a settlement to be permanent, allowing for ephemeral urbanity
(dense but temporary settlements). While density-based approaches to classifying human
settlements based on gridded population data are conceptually robust, such as the
methodology adopted by the United Nations in 2020 for cross-national comparison, we
present an alternative measurement approach that is more closely aligned with our
phenomenological understanding of urbanity.
Keywords: urbanity, urbanism, urbanization, city, human settlement classification
1
Introduction
The concept of an “urban place” is ancient and ambiguous. It is pervasive in popular culture
and public discourse, it is a foundational concept and site of study in many academic subdisciplines, and it is one of the most enduring statistical categories deployed by researchers
and policy makers. Yet there is no consensus definition of what makes a place “urban.” Every
country in the world has a unique “urban” definition for the purposes of statistical
classification (Cohen, 2004; Buettner, 2015). Moreover, the central concepts in these definitions
can change over time: in 2021 the United States Census Bureau changed a long-standing urban
definition from one emphasising population characteristics in census tracts to one
emphasising residential structure density (USCB 2022). This lack of consensus and consistency
powers a rich theoretical literature, but also complicates empirical research and has concrete
policy implications.
If we are to develop an accurate understanding of the ongoing global urban transition—one
of the most profound geodemographic phenomena in human history—we need a
conceptually robust and globally consistent definition of urbanity for measurement and
comparison. This was a key motivation for the development of a new global classification
system for human settlements, known as Degree of urbanisation (DEGURBA), which was
formally adopted by the United Nations in 2020. Developed by the European Commission,
the DEGURBA system uses gridded population data and simple population size and density
criteria for human settlements classification.
However, this approach is controversial, with some suggesting demographic criteria alone
are insufficient to capture the nature of urban places (Angela et al 2018). We argue that debates
about empirical definitions of “urbanity” arise from conceptual confusion about what place
attributes are distinctly “urban.” Broadly speaking, existing definitions used for human
settlements classification incorporate some combination of geodemographic attributes (such as
population size and density), socioeconomic or political attributes (such as labour market
structure or political institutions), or the physical characteristics of the built environment (such
as presence of public infrastructure). These are related: geodemographic attributes affect the
shape and structure of socioeconomic and political phenomena, which often unfold in (or
reshape) built environments. Nevertheless, they are conceptually distinct types of attributes.
Attending to this distinction allows us to differentiate between cities and urban places. While
cities are the physical environments most commonly associated with urbanity, we argue that
not all cities are urban, and some urban places do not have the physical or political
characteristics traditionally associated with cities. From this perspective, empirical efforts to
“delineate urban areas” using data on building footprints are better understood as efforts to
identify cities rather than urban places (e.g. Taubenbock et al. 2012, Usui 2019, Montero et al.
2021, Bellefon et al. 2021, Arribas-Bel et al. 2021). In contrast, we argue that urbanity arises
from bodies, not bricks.
These distinctions are important, because the geodemographic, socioeconomic, and physical
attributes of cities and urban places have often been conflated in urban research, undermining
the analytic value of the geographic concepts of the city and urbanity. Louis Wirth, in his
highly influential essay Urbanism as a Way of Life (1938), asserted that “a city is a relatively
large, dense and permanent settlement,” which is a strictly geodemographic definition. But
he went on to highlight the “complex of traits which makes up the characteristic mode of life
in cities” (pg. 7), such as economic diversification and exchange, superficial human relations,
social heterogeneity and segregation, and the emergence of complex and overlapping social
group structures. Similarly, in his classic essay The Urban Revolution (1950), V. Gordon Childe
proposed a list of 10 criteria for identifying ancient cities, including their demographic size
and density, but also the presence of a wide range of non-agricultural specialists and “truly
monumental public buildings.” More recently, Scott (2021) has argued that “’urban’ refers
2
both to the city as a concrete entity and to urbanisation as the process by which the city is
generated and socially reproduced” (2) and that “the division of labour is fundamental to the
genesis and durability of cities” (3). In each case ”the urban” and “the city” are conflated and
used with reference to a host of social and physical characteristics. Both terms become
conceptual vessels for describing complex cultural, socioeconomic, and political
developments—assemblages of versions of modernity.
To confuse matters further, scholars have used these fuzzy concepts to describe conditions in
non-urban places. Wirth argued that cities foster modernity, but they do not confine it
geographically: “the urban mode of life” (a socioeconomic phenomenon) “is not confined to
cities” (a type of geographic locale). More recently, Brenner and Schmid (2014) have argued
that there is no longer an outside to ‘the urban’ given the planetary-scale consequences of
contemporary urban living. Urban places exert cultural, economic, environmental, and
political influence far beyond their functional or administrative boundaries (cf. Jacobs, 1984;
Sayer, 1984; Brenner and Schmid, 2014; Shin, 2017). While theoretically stimulating, conflating
cities and urbanity with modernity or capitalist globalisation undermines the geographic
specificity of these concepts and their application to empirical research. If ‘urban places’ have
engendered a ‘way of life’ that can be found in non-urban places, this way of life cannot be
used to empirically differentiate between urban and non-urban places. Yet most people would
agree that such places exist and are different. We attempt to disentangle this conceptual
morass by making clear distinctions between the geographic concepts of urbanity and ‘the
city’, and between their geographic nature and their socioeconomic or political implications.1
Our focus, then, is on identifying the core defining characteristics of urbanity rather than on
defining “the city.” Following key principles of concept formation, we aim to define urban
places in a way that is (1) familiar, (2) offers both theoretical and practical utility, (3) clearly
differentiates urban places from non-urban places, and (4) does so with reference to
observable attributes (Gerring, 1999). Like Scott (Scott, 2017, p. 23), we seek a “disciplined
minimal concept,” but in our case we seek to define urbanity rather than “the city.”
We focus very narrowly on identifying the attributes that are both necessary and sufficient for
a place to be “urban” rather than theorising about social processes that may arise there or
reach far beyond a particular location. We intentionally avoid the word ‘urbanism’ and its
sociological (and ideological) connotations. Approaching this challenge from a comparative
international perspective, we recognise that there are many ways of life to be found in urban
places—and many types of urban places (Parnell and Pieterse 2015). Consequently, a singular
grand theory of “the urban” is problematic. Yet there is scope for a universal definition of
urban as an adjective for the purposes of geographic classification, statistical measurement,
and comparative research. Indeed, clear criteria for classifying human settlements can help
support an appreciation and systematic analysis of this diversity.
Building on a series of thought experiments motivated by diverse existing approaches to
human settlement classification, we argue that urbanity is experienced wherever there is a
large concentration of human beings in a geographic place for a sustained period of time. In
practical terms this means that urbanity can be measured with reference to the demographic
characteristics of places alone. Urban places are where the people are. This conceptualisation
echoes Wirth’s geodemographic definition, but makes space for ephemerality: the urbanity of
a place can wax and wane, on timescales shaped by social or physical factors that mediate
urban ways of living. Our emphasis on physical interaction and co-presence reveals the
shortcomings of those who see the city purely as a networked place of social interaction. We
can see clearly now, in the 21st century, that social interaction and networks can be abstracted
from places and exist in wholly disembodied forms. Physical copresence often yields distinct
forms of social contact.2
3
Our conceptual approach resonates with the DEGURBA approach adopted by the UN
Statistical Commission in 2020 to facilitate international comparison. However, while we
make the case for the primacy of demographic factors in the definition of urbanity, we suggest
that current approaches relying on strict population density thresholds are problematic. It is
now technically possible to measure urbanity along a continuous spectrum and describe
settlements and settlement systems in more nuanced ways than traditional measures such as
rural/urban ratios or primacy measures. Improvements in both measurement and
computation make far more sophisticated analysis possible (Hugo, Champion and Lattes,
2003; Schroeder and Pacas, 2021).
To illustrate how we might better align our phenomenological understanding of urbanity
with empirical measurement, we introduce a novel approach that tackles the perennial
boundary problem in geographic research. Rather than classify places based on population
density—which requires a boundary to be imposed “from above” – we introduce a
“population proximity index” that requires no such imposition. Instead, it is a continuous
measure of population concentration in a place that reflects the “catchment” area required to
reach a minimum population threshold at any point on the planet. Places with small
catchments have high degrees of urbanity; places with large catchments have low levels of
urbanity. Broadly speaking, we call this concept “population proximity.”
Some scholars resist discrete geographic classification of places, emphasising instead the
socioeconomic, political, and ecological processes that link human settlements to each other
and to uninhabited landscapes (e.g. Brenner and Schmid 2014). We contend that there are
urban places and rural places, and these are observably different types of places—just as
valleys and mountains are distinct features of the landscape, even if the precise dividing line
between the two is subject to debate. The fates of hamlets and villages and towns and cities
may be deeply intertwined, and the socioeconomic and ecological consequences of urban
settlements may be planetary in nature, but these observations do not justify a rejection of
human settlement classification altogether. If anything, these observations bolster the
argument for more sophisticated and relevant approaches to human settlements analysis built
on firm conceptual foundations.
Decoupling geodemographics from social processes
Wirth’s essay is a useful foil for thinking through the problems of eliding the classification of
places with observations about the social phenomena within them. In his own words, Wirth’s
goal was to “discover the forms of social action and organization that typically emerge in
relatively permanent, compact settlements of large numbers of heterogeneous individuals”
(pg. 9). It was an explicit attempt to generalise the social consequences of urban living. Wirth
therefore understood urbanization to be the “development and extension” of the “complex of
traits which makes up the characteristic mode of life in cities” to other types of settlements—
not merely growth in the share of a region’s population living in “cities in the physical and
demographic sense” (pg. 7). In this understanding, Wirth’s theory of urbanism is an historical
theory of broad social change, building on the work of scholars like Durkheim and Weber.
Wirth argued that classifying places based upon arbitrary population thresholds obscured
analysis of important social processes that are not geographically confined.3 This is sensible:
an arbitrary population threshold is unlikely to define when urban social processes start or
stop. Further, there is no inherent logical problem with using Wirth’s theory of urbanism to
understand the consequences of large numbers of people living in cities defined in the
“physical and demographic sense.” However, it is problematic to use this theory of urbanism
to classify places because the social consequences of urbanism can arise in very different
circumstances than the ones Wirth analyses. Unfortunately, the power of Wirth’s critique has
4
led many scholars and statisticians to do exactly this, inverting Wirth’s historical theory to
classify geographic places with peculiar results.
We can see this in the diverse criteria that have been used by national statistical agencies to
classify human settlements as either rural or urban. While many countries use demographic
criteria consistent with Wirth’s foundational geodemographic definitions (population size
and density), many also include criteria that reflect a holistic or historical understanding of
urbanization and urbanity. Broadly speaking, these include administrative status (e.g.
“Localities proclaimed as urban”), economic characteristics (e.g. labour market structure), or
the presence of “urban characteristics” in the form of physical infrastructure or amenities (e.g.
paved roads, electricity)(Buettner, 2015). 4 According to an analysis of the United Nations’
World Urbanization Prospects 2011 Revision, 121 countries use a single criterion to classify a
settlement as rural or urban, with 64 using administrative criteria, 48 using demographic
criteria and nine using ‘urban characteristics’ (ibid). A further 65 countries use two criteria,
while a further 19 use three or four criteria (ibid). We consider the appropriateness of each of
these classes of criteria in turn.
The urban as a space of special administrative status
As noted by Wirth, “the city, statistically speaking, is always an administrative concept” (pg.4).
Indeed, administrative status is the most common criterion for urban settlement classification
among the 121 countries that use a single criterion. This can yield counterintuitive results. For
example, the city of St David’s in Wales registered a population of just 1,751 residents in the
2021 census—a population smaller than many settlements without “city” status in the UK. Its
status as a city is due to the presence of a cathedral and its historical role as a site of Christian
worship. The settlement was stripped of city status in 1886 but it was re-designated as a city
in 1994. Yet it is doubtful that visitors to the bucolic St David’s would describe the settlement
as urban.
Conversely, a place may be identifiably urban but not a city. Clear examples can be found in
India. There is a category of settlement referred to as Census Towns, which are settlements
that have grown to a minimum population size of 5000 living at a minimum population
density of 400 people per square kilometre and with at least 75 per cent or more of the male
population engaged in non-agricultural activities (Mukhopadhyay et al., 2016). These are
classified as ‘urban’ for census purposes but retain rural administrative status, which has
important implications for resource allocation . More awkwardly, the ‘village’ of Rahri, which
is classified as rural, registered a population of 36,569 in the 2011 Indian census—much larger
than most ‘urban’ census towns.
These cases clearly illustrate the ontological distinction between urbanity, which we argue has
a universal character, and “city-ness,” which is highly contextual. Simply put, a place need
not be urban to be a city, and an identifiably urban place may not necessarily be a city.
Administrative status is too arbitrary and politically contingent to rely upon for a durable
definition of urbanity.
The urban as a labour market
Many statistical and social scientific definitions of urbanity refer to the characteristics of local
labour markets. As noted above, settlements in India must have at least 75 per cent of the male
population working in non-agricultural jobs. Over 30 countries employ similar thresholds as
part of a multi-criteria classification system (Buettner 2015) as do many scholars (Wirth 1938;
Shin 2017; Angel et al. 2018; Scott 2021). This is problematic when it leads to a classification
system that runs counter to the lived experience of place.
5
For example, workplace data is used by the OECD to build “functional urban areas,” linking
the self-evidently densely populated urban “core” to “peripheral” places where people
commute into the city to work (Dijkstra et al., 2013). This is reasonable when delineating the
space of urban functions, but these places are not “urban” in a geodemographic sense. These
places may be functionally integrated and one (or more) may be urban, but this does not mean
that all places within a functional urban region are urban themselves. Indeed, it is precisely
their less-than-urban character (i.e. suburban or exurban) that some find attractive. This
confusion propagates when “function” is not clearly specified; the 2020 US Census’s revised
definition of “urban” now merges places together where commuting is “50 percent in at least
one direction,” (USCB 2022, p. 16712), bringing relatively small (2,500-person) commuter
villages into the urban fold. This is no longer a “functional” area surrounding an urban core,
but rather a definition of the “urban” place itself. Technological and social changes that have
made such “commuter villages” possible will continue to heighten the contradictions between
urban places and the economic characteristics of functional urban areas. Urban
neighbourhoods and commuter villages feel different even when their labour market
characteristics are very similar—a difference that we argue arises primarily from distinct
geodemographic contexts.
In these cases, we find the roots of urban economies reaching deep into rural landscapes—a
phenomenon Wirth and many others before and since have highlighted. But it does not follow
that these commuter communities are therefore urban unless we entirely discard
demographic criteria and ignore human perception and experience. Some theorists may be
comfortable abstracting urbanity from geodemographics; most people would likely find this
unintuitive. Given that we seek a familiar understanding of urbanity, we argue that economic
characteristics like employment sector or commuter balance are neither necessary nor
sufficient to classify a place as urban or rural. These “functional” characteristics situate places
in relation to geographically extensive economic systems, but they do not capture relative
urban character of a settlement.
The urban as a built environment
This brings us to the final (non-demographic) class of criteria commonly used to classify places:
the presence of physical infrastructure and amenities, such as piped water, paved roads and
medical or educational facilities. Forty-four countries use such “urban characteristics” as part
of their settlement classification system (Buettner 2015). In popular culture, urban places are
frequently represented as dense built environments with large, multi-story towers and
congested streets—places like New York, London, Tokyo, Shanghai, Lagos, Sao Paulo or
Mumbai. We imagine buildings and infrastructure forming a larger-than-human built
environment with a sense of monumentality. And again, these built environments often host
urban places. But there are abundant examples of places with highly developed infrastructure
that would not be classified as urban, as well as urban places that lack such infrastructure.
Consider the ancient city of Tikal in Guatemala, which reached a population of 62,000 around
AD 700 (Wright, 2012). Today the temples and squares and water infrastructure are a tourist
attraction. But is it urban? With no inhabitants, Tikal is a physical artefact—perhaps a city but
surely not an urban place. If one accepts this line of reasoning, the same could be said of socalled “ghost cities” in China (Sorace and Hurst, 2016) or any lightly-populated “urban
laboratory” like Paolo Soleri’s Arcosanti (Evans et al. 2016). These are human modified
landscapes, but they are essentially devoid of the complex and intensive human activity we
recognize as urban. The presence of buildings and infrastructure to support human habitation
may render them suitable for classification as “cities” but they have none of the social potential
of a crowded place. And, while we suggest ephemeral urbanity to explain how these places
6
might easily become urban, their existence as bundles of infrastructure does not make them
urban outright.
In contrast, the sprawling “slums” or informal settlements of cities in many low- and middleincome countries often lack permanent buildings and the kinds of water and sanitation
infrastructure that are associated with ancient and contemporary cities alike. Yet they have
the ‘buzz’ that comes from the concentration of human interactions that arise from population
density—the type of buzz characteristic of urban places (Storper and Venables, 2004).
Similarly, refugee camps are often characterised by high population densities, impermanent
structures, and the absence of infrastructure. Yet the concentration of people in large camps
gives rise to many of the very phenomena that social scientists and statisticians associate with
urban living, such as socioeconomic specialisation and exchange and the emergence of land
and rental markets. For example, the Zaatari refugee camp in Jordan, which was established
in 2012 to support refugees from the Syrian war, has become home to roughly 80,000 people
(Dalal, 2015). It has a lively souk fuelled by both humanitarian aid and an informal labour
market and “all kinds of business one expects to see in a city” (ibid, 271). The camp has
evolved into something that feels distinctly like a de facto city despite the absence of major
infrastructure or permanent structures.
Urbanity beyond the legal, economic, and (infra)structural city
People make places urban. Sometimes these places have monumental buildings and
infrastructure ferrying workers to jobs regulated by special legal or administrative
arrangements. But there are many urban places in the world that have none of these things
yet are experienced as urban places by residents and visitors alike. Decoupling the
geodemographic characteristics of a place from its political, socioeconomic, or technological
characteristics throws into relief the essential characteristic of urbanity. The constant that
unites New York City, a Mumbai slum and a Jordanian refugee camp is the geographic
concentration of people in space. In all these places the geographic concentration of people—
or mass corporeal copresence—has common consequences and stimulates diverse emergent
social phenomena.
Corporeality, strangers, and permanence
Wirth highlighted the “coincidence of close physical contact and distant social relations” (1)
as a defining aspect of urban living. Similarly, Iris Marion Young defined the essence of “city
life” as “the being together of strangers” (Young 1990). Here we find the truly distinguishing
attributes of urbanity. It is the sight and sound of strangers all around; the possibility of a
physical encounter with an “other”, unfamiliar person. It is the geographic concentration of
human bodies needing food and water, and excreting waste, that creates an urban context. It is
a geographic concentration of people—not buildings—that make places urban.
Disembodied forms of human congregation (e.g. on social media platforms) do not replicate
physical co-presence. Non-corporeal copresence and contact is distinctive by virtue of its
technological mediation. In some cases, digital interaction complements physical interaction
instead of supplanting it (Craig et al. 2017; Kujath 2011; Sinai and Waldfoegel 2004). But there
are systematic differences in the nature of interpersonal contact online versus in person.
Research has shown that experiences are amplified when shared when others, and spatial
proximity is an important mechanism of joint experience (Boothby et al 2016). Online
interactions offer fewer nonverbal cues and greater anonymity between individuals
(Lieberman and Schroeder 2020). And there is evidence that groups work differently online
versus in person. Research at Microsoft during the COVID-19 pandemic showed that
7
increased online work reduced interaction between groups, resulting in “static and siloed”
communications (Yang et al. 2022), while a recent study across five countries found that teams
relying on videoconferencing produce fewer creative ideas than ones working in person
(Brucks and Levav 2022). At a very fundamental level, disembodied interactions do not have
the potential for immediate physical contact—for physical violence, intimacy, or disease
transmission.
Urban places facilitate social interaction (like the internet), but the internet cannot replicate
the effects of physical proximity inherent in urban places. Thus, corporeal copresence and its
environmental effects is distinct from copresence in general and is necessary for a uniquely
urban experience. It isn’t just the enhanced possibility of contact with strangers, but the
enhanced possibility of physical contact that makes urban places distinct from non-urban
places.
The presence of strangers is also a key feature of urbanity. Wherever people congregate in
sufficient numbers there will be strangers. This fact is physically determined: it is not possible
to maintain meaningful interpersonal relationships with an infinite number of people because
meaningful relationships require time and attention to maintain. Empirical research suggests
that people generally maintain between 100-300 “meaningful” relationships (Dunbar, 1992;
Gonçalves, Perra and Vespignani, 2011). As a result, from the perspective of any individual,
the share of strangers in a group increases with group size.
For example, if we take the most common estimate of the number of people any individual
can maintain meaningful relationships with (150), the share of strangers for any individual in
a community of 1,500 is 90%. If the relationship threshold is doubled to 300, a “stranger share”
of 90% is achieved in a group of 3000 individuals.
Figure 1 further illustrates of how population size relates to the share of strangers in a place.
In small populations, each additional member makes a sizeable contribution to the stranger
share (over any given threshold). But the effect rapidly diminishes with scale: in large
populations each additional stranger makes a negligible contribution. In a small town, every
new resident may be noticed; in London, New York or Lagos new arrivals are unseen by the
vast majority of residents.
This thought experiment helps to provide an answer to the question “how large is large?” that
follows naturally from Wirth’s geodemographic definition of an urban settlement as a “large,
dense and permanent settlement.” If we accept that physical copresence with strangers is a
defining characteristic of urbanity, we can calibrate the intensity of urbanity in relation to the
share of strangers in one’s immediate vicinity. It is notable that minimum population
thresholds have long been used by statistical agencies in classifying urban places (Johnston,
1980, p.13, see also Truesdell (1949) for the US from 1874-1949) and now range between 200
and 50,000 (Bandyopadhyay and Green 2018; Buettner 2015). The UN’s harmonized
classification system (DEGURBA) identifies high density settlements of 50,000 people or more
as urban centres. Even with a generous “relationship threshold” assumption, over 99% of
people in such a settlement will be strangers to one another.
Figure 1 | The share of strangers as a function of group size
8
It is not our contention that there is a “natural” threshold at which a group grows into an
urban community, but rather that group size is a critical variable because it conditions social
context in a universal fashion (i.e. through the relative presence of strangers).5 This generates
demand for social institutions to manage relations between strangers (pace Wirth), but there
is no automatic solution to this challenge at any density of habitation.
What of permanence? Must a congregation of human beings be permanent to qualify as urban?
There is something intuitive about this idea, but it quickly breaks down when we are asked
to define permanence. For example, Black Rock City is “a temporary metropolis” that springs
up once a year in the Nevada Desert as part of the Burning Man festival (burningman.org/allevents/ - accessed 03 July 2023). Putting aside whether there is enough infrastructure for it to
be considered a city or “metropolis,” it is certainly a “notable urban and geographical
experiment” (Rohrmeier and Starrs, 2014; 169). For one week every year, tens of thousands of
people congregate to live together in a temporary settlement with a distinctly urban character
(see Figure 2). It has even been held up as a potential model for managing rapid urban growth
in low- or middle-income countries (Keil, 2021). We would argue that Black Rock City is an
urban place when it exists, but that the quality of urbanity evaporates from the site when the
revellers return home. It is an exemplar of ephemeral urbanity.6
Figure 2 | Black Rock City
9
Source: REUTERS/Jim Urquhart
Measuring urbanity by inverting the boundary question
Spiro Kostof argued that cities are “where a certain energized crowding of people takes place.
This has nothing to do with absolute size or with absolute numbers; it has to do with
settlement density” (1991, 37). As he observes, for most of our approximately 6000-year urban
history very few towns or cities had populations greater than 10,000. Their identifiable
urbanity was essentially a function of population density (ibid). But how dense is dense?
Because density is a relationship between population sizes and geographical extents, any
urban definition based on density requires us to deal with the fact that density is an “intensive”
property, dependent on (a) how many people are (b) within a boundary of measurement.
Taking each in turn, there is no obvious or natural threshold population size at which a human
settlement or region becomes urban. A population of 50,000 living in an area 50,000 km2 is not
intuitively urban; the same population living in 10 km2 likely is. Thus, it is often the
measurement boundary (and not the size of a given population) that determines the “urbanity”
of that population given that this ultimately influences measured density (pace Kostof).
Traditionally, statisticians have used administrative or enumeration boundaries to make such
calculations. But there is no global consensus on how to draw urban boundaries; the choice is
generally pragmatic, driven by the application at hand and available data (Storper and Scott,
2016).
For the purposes of global comparison, the DEGURBA classification system uses a simple and
transparent approach to deal with the non-comparability of statistical definitions across
countries. Drawing on gridded population data, geographic regions are constructed and
classified as “urban centres” if there are one or more contiguous cells that contain a minimum
of 50,000 people living at a density of 1500 inhabitants per square kilometre; a region is
classified as an “urban cluster” if it contains a minimum of 5000 people living at a density of
300 inhabitants per square kilometre. Collections of regular 1km2 grid cells provide the
necessary boundaries to calculate density: the outer edges of contiguous cells form the
“natural” boundary of an urban centre or cluster, irrespective of local administrative
geographies. However, this approach to imposing de facto urban boundaries is both arbitrary
and rigid. As Angel et al. (2018) observe, the DEGURBA classification thresholds do not
10
necessarily match the lived realities of both high-density rural communities or those living on
the fringes of cities. A more nuanced hierarchy of thresholds could be used to improve
representations of the different ways of living that are condensed into “urban” classifications7,
but the the resulting boundaries of the urban places themselves will always be an artificial
imposition only loosely tethered to human experience.8
The arbitrariness of these boundary problems and their effect on aggregate estimates is wellknown in geographical research (Openshaw & Taylor, 1979)—the “mythical beast” of the
urban boundary is “a little difficult to capture” (Hall, 2007), even with new data and stronger
methods. This is because the boundary issue is a theoretical, not empirical issue: any boundary
is not only fuzzy, it is local: the boundaries for a city are contingent on a wide variety of local
physical factors, and the odd and uneven surface of our Earth can yield an extremely large
diversity of unusual corner cases where places may feel urban, but do not meet arbitrary
inclusion/exclusion thresholds at the urban edge.
Inverting the boundary problem
There is, however, an alternative way to use the concept of density to measure urbanity that
does not require the urban geography of a place to be imposed from above. People do not
experience districts or enumeration blocks or grid cells in a visceral way; we feel the presence
of other people through sight, sound, smell, and touch. We can feel the energized crowding
or the absence of it. Indeed, using data from the American Housing Survey, a study by the US
Census Bureau found that the single best predictor of whether a respondent identified their
area as rural, suburban, or urban was the population density of their area (Bucholtz, Molfino
and Kolko, 2020). Thus, the perception of urbanity is intrinsically linked to individuals’
experiences of population concentration.
Our suggestion is therefore to invert the boundary problem, borrowing from sociological and
community psychology work on ‘egohoods’ (Hipp and Boessen, 2013). Instead of focusing on
density, we can instead measure urbanity with an alternative indicator of population proximity:
the distance (or time) one would have to travel until some specific threshold (say, 99%) of the
people encountered in that generalised activity space (Spielman and Singleton, 2022) are
strangers.
Practically speaking, we can construct a “population proximity index” (PPI) as a proxy for
urbanity as d, the distance you would have to travel (in any direction) from a given location
to reach some catchment population, k.9 This distance is effectively the urbanity value of a
location. If the area required to capture some minimal population is very large, that location
is not very urban; if it is a small area, it is a very urban location. Instead of measuring
population density within a defined set of boundaries, we are measuring the boundary
required to meet a certain population threshold, thereby “inverting” the boundary problem.
In other words, the geography of urbanity emerges “organically” from this approach, offering
a more nuanced representation of urbanity as experienced. This can be seen in Figure 3, which
visualises differences in population density and population proximity for three case study
cities: Nairobi, Kenya; Sao Paulo, Brazil; Sydney, Australia.
11
Figure 3 | Comparing population density and population proximity in three cities.
Population data are from WorldPop 2020 (constrained). Population Proximity was calculated with a
medium population threshold of 3000 people. In the disagreement pane, red indicates strong
disagreement, while yellow indicates weak disagreement.
The differences between standard gridded population (left column) and population proximity
are not dramatic, but significant at the margins—particular in per-urban areas and urban
fringes. This is illustrated by the central column of panels, which shows disagreements
between the two for each city (red is strong disagreement; yellow is weak). The greatest
differences occur around the urban fringes. This makes sense: in areas with high population
counts in cells, the population proximity index will be small (i.e. highly urban). In effect, the
PPI is censored at the lowest values (i.e. highest degrees of urbanity). Consequently, it is areas
12
of moderate and variable population density that see the greatest disagreements. Among
these cases, Sydney is the starkest example. The extensive suburbs to the north and southwest
of the city center appear substantially less urban with the PPI than a measure of population
density alone might suggest.
This is in part due to the way in which the PPI embeds information about geodemographic
context. In a standard population grid, every cell contains an independent measure of
population, with no reference to ‘neighbourhood’ context. By contrast, the PPI reflects
demographic information about the wider neighbourhood, offering a more nuanced measure
of context at any given location.10 In this sense, our approach resonates with the settlement
classification system recently developed in the US by Schroeder and Pacas (2021), which uses
population size and concentration as key dimensions for measuring urbanity across a
spectrum. In this system, one can find a distinctly rural-like (i.e. low density) settlement within
a wider (urban) metropolitan region. However, our proposed approach does not require
regionalised data and is therefore more globally applicable.
Figure 4 illustrates the effects of changing the population threshold of the PPI, using the case
of Sydney and thresholds of 1,500, 3,000 and 5,000. There is little change in the urban core, but
as the threshold increases peripheral areas appear increasingly less urban, with a notable loss
of urban “hotspots” in the outer areas. Increasing the PPI threshold throws into sharper relief
the most “urban” areas of a large urban region.
Figure 4 | Population proximity over three thresholds in Sydney, Australia.
Finally, Figure 5 shows how the relationship between population density and PPI changes
with settlement size (measured in terms of population size). Here, the cell-level correlation
between density and PPI has been calculated for 49 urban areas. The correlation coefficients
are then plotted against city size (log of city population). This “correlation of correlations”
shows us that (a) the cell-level correlation is consistently negative, as we would expect
(smaller PPI = greater urbanity), and (b) that the strength of the correlation increases with city
size. This is consistent with the evidence in Figures 3 and 4, which show the areas of greatest
disagreement and sensitivity to population threshold are those of moderate population
density.
13
Figure 5 | Cell-level correlation between PPI & density in 49 cities. The Figure shows the rank
correlation between population proximity and density at cell-level in a global 49 city sample across
three population thresholds. Note the cell-level correlation is negative because dense places tend to
have small PPI values (i.e. exhibit greater urbanity) and this negative correlation increases with city
size.
This approach to measuring urbanity has some practical and conceptual advantages over
population density. Practically, it offers a high-resolution “organic” representation of
urbanity within any given city or region, at any moment in time, or even at different times of
day. As argued above, urbanity-as-experienced ebbs and flows with the demographic tides.
Conceptually, the population proximity index re-centers the physical co-presence of strangers
as the animating concept for urbanity and makes explicit what density makes implicit: the
geography of urban experiences. Measures of population density must treat the geography of
populations as a ‘known’ fixed container in which it makes sense to count people—invariably
enmeshing density with intractable geographic questions about scale and the modifiable areal
unit. Population proximity resolves this issue by treating scale (i.e. distance to the catchment
population) as an outcome, rather than as an assumption required implicitly by the measure.
The benefit of his approach is scalability and comparability, and further metrics could be
developed from it to provide comparative statistics, such as average population proximity for
whole cities or PPI variability across urban regions. But this approach doesn’t capture the
nuance of actual human activity spaces. As Stülpnagel et al. (2019) demonstrate, simple circles
rarely reflect the true geography of individual activity and hence the social context they may
experience. Alessandretti et al (2020) show that individuals have multiple scales of mobility,
which can be significantly impacted by gender, transport infrastructure and settlement type.
People experience similar places differently, which can substantially alter their experience of
urbanity—their regular exposure to large numbers of strangers. Our simple cartesian
approach glosses over much more complex social dynamics. Nevertheless, it offers a
pragmatic approach to evaluating urbanity in a relatively nuanced way across spatial scales
with minimal data, and can easily incorporate richer approaches to modelling activity spaces
when data is available.
Conclusion
The question posed in the title of this paper signals our intentional treatment of urban as a as
an adjective. Rather than seek to define the urban or the urban condition (nouns) we have made
a case for defining urban as a geodemographic attribute—one that can be found in a wide
range of places where large numbers of people congregate for sustained periods of time.
This perspective sets our case apart from those such as Wirth (1938) and Scott (2021). In an
effort to define the urban way of life or the city, these scholars tend to over-specify the
14
consequences of urbanity, while many others make unnecessary assumptions about the
causes of urbanity or conditions under which it arises. An urban place only emerges where
there is sufficient surplus to support a dense concentration of people (Childe 1950), but this
surplus can be from local production, trade, humanitarian aid or some other mechanism of
acquisition. And while living in close proximity creates similar challenges wherever it
occurs,11 there are many ways these challenges can be and are addressed. Diverse urbanisms
emerge from the shared context of urbanity. There is nothing deterministic about the social
and economic consequences of human congregation. To quote (Merrifield, 2013):
The urban is nothing in itself, nothing outside dynamic social relations, a coming together of
people. As long as human beings can come together, as long as separation can be resisted, there
is always a possibility of encounters between people. (p. 916)
It is the possibility of physical encounter—and frequent encounter with strangers in
particular—that gives a place an urban character. There is no specific form or function to
urbanity; it is merely a geodemographic context ripe with social potential. This means that
urbanity is, at its core, ephemeral: it emerges as well as dissipates. This is experienced through
the daily rhythm of concentration and de-concentration in cities with large commuter belts.
As downtowns swell in population during working hours, they change in character—they
become more urban. As they empty at night, they become less urban. This has significant
implications for planners and policy makers who need to account for de facto urbanity – how
population density is actually experienced in places – rather than residential density (Boeing
2018)and highlights the importance of recent innovations in estimating and analysing daytime
populations (ibid; Moro et al. 2021; Xu, 2021). Residence remains an important aspect of
urbanity as it is experienced but should be treated as one aspect among many. Similarly, we
have argued that “permanence” should not be considered necessary for a place to be
considered urban. Just as urbanity ebbs and flows in large cities, it can emerge and dissipate
in places outside of cities, with festivals and refugee camps as prime examples.
Our proposed approach to measuring of urbanity, inspired by the corporeal essence of urban
places, inverts the traditional approach to geodemographic place classification by doing away
with pre-defined boundaries from administrative geographies or grid cells. Instead, variable
population catchment areas are used to encode information about the geodemographic
contexts of locations, and this can be done at multiple spatial and temporal scales. This
globally applicable approach offers a nuanced way of representing urbanity as experienced
by people.
15
References
Angel, S., Lamson-Hall, P., Guerra, B., Liu, Y., Galarza, N., & Blei, A. M. (2018). Our Not-SoUrban World. Working Paper No. 42, The Marron Institute of Urban Management, NYU.
http://www.thisisplace.org/i/?id=0150beca-e3f5-47e0-bc74-9ccc5ef1db8a
Batty, M. (2013) The n ew science of cities. Cambridge, MA: MIT Press.
Bandyopadhyay, S., & Green, E. (2018). Urbanization and mortality decline. Journal of
Regional Science, 58(2), 483–503. https://doi.org/10.1111/JORS.12375
Bhagat, R. B. (2005). RURAL-URBAN CLASSIFICATION AND MUNICIPAL
GOVERNANCE IN INDIA. Singapore Journal of Tropical Geography, 26(1), 61–73.
https://doi.org/10.1111/J.0129-7619.2005.00204.X
Boeing, G. (2018). Estimating local daytime population density from census and payroll
data. Regional Studies, Regional Science, 5(1), 179–182.
https://doi.org/10.1080/21681376.2018.1455535
Boothby, E. J., Smith, L. K., Clark, M. S., & Bargh, J. A. (2016). Psychological distance
moderates the amplification of shared experience. Personality and Social Psychology
Bulletin, 42(10), 1431-1444.
Brenner, N. and Schmid, C. (2014) The ‘urban age’ in question. International Journal of Urban
and Regional Research, 38(3), pp. 731–755. Available at: https://doi.org/10.1111/14682427.12115.
Brucks, Melanie S., and Jonathan Levav. 2022. ‘Virtual Communication Curbs Creative Idea
Generation’. Nature 605 (7908): 108–12. https://doi.org/10.1038/s41586-022-04643-y.
Bucholtz, S., Molfino, E. and Kolko, J. (2020) The Urbanization Perceptions Small Area
Index: An application of machine learning and small area estimation to household
survey data. Working Paper, U.S. Department of Housing and Urban Development, US
Census Bureau. Available at:
https://www.huduser.gov/portal/sites/default/files/docs/UPSAI_forWeb.docx
Buettner, T. (2015). Urban Estimates and Projections at the United Nations: The Strengths,
Weaknesses, and Underpinnings of the World Urbanization Prospects. Spatial
Demography, 3(2), 91–108. https://doi.org/10.1007/s40980-015-0004-2
Bureau of the Census, D. of C. (2021) “Urban Areas for the 2020 Census - Proposed Criteria,”
Federal Register. United States: Bureau of the Census, Department of Commerce, pp.
10237–10243.
Childe, Gordon V. (1950) The urban revolution. The Town Planning Review, 21(1), pp. 3-17.
Cohen, B. (2004) Urban growth in developing countries: A review of current trends and a
caution regarding existing forecasts. World Development, 32(1), pp. 23–51. Available at:
https://doi.org/10.1016/j.worlddev.2003.04.008.
Craig, Steven G., Edward C. Hoang, and Janet E. Kohlhase (2017) Does Closeness in Virtual
Space Complement Urban Space? Socio-Economic Planning Sciences, Digital Support
Tools for Smart Cities, 58 (June 1): 22–29. https://doi.org/10.1016/j.seps.2016.11.002.
Dalal, A. (2015) A Socio-economic Perspective on the Urbanisation of Zaatari Camp.
Migration Letters, (3), pp. 263–278. Available at: www.migrationletters.com.
Dijkstra, L., H. Poelman and P. Veneri (2019) The EU-OECD definition of a functional urban
area. OECD Regional Development Working Papers, No. 2019/11. OECD Publishing, Paris,
https://doi.org/10.1787/d58cb34d-en.
16
Dijkstra, Lewis, Aneta J. Florczyk, Sergio Freire, Thomas Kemper, Michele Melchiorri,
Martino Pesaresi, and Marcello Schiavina. 2021. ‘Applying the Degree of
Urbanisation to the Globe: A New Harmonised Definition Reveals a Different Picture
of Global Urbanisation’. Journal of Urban Economics, Delineation of Urban Areas, 125
(September): 103312. https://doi.org/10.1016/j.jue.2020.103312.
Dunbar, R.I.M. (1992) Neocortex size as a constraint size in primates on group ecologically,
Journal of Human Evolution, 20, pp. 469–493.
Evans, J., Gabriele S., and Luke, K. “The Glorious Failure of the Experimental City.” In The
Experimental City, 218–35. Oxford, England: Routledge, 2016.
Gerring, J. (1999) What makes a concept good? A criterial framework for understanding
concept formation in the social sciences. Polity, 31(3), pp. 357–393. Available at:
https://doi.org/10.2307/3235246.
Gonçalves, B., Perra, N. and Vespignani, A. (2011) Modeling Users’ Activity on Twitter
Networks: Validation of Dunbar’s Number. PLOS ONE, 6(8), p. e22656. Available at:
https://doi.org/10.1371/JOURNAL.PONE.0022656.
Hipp, J. R., and Boessen, A. (2013). "Egohoods as waves washing across the city: A new
measure of “neighborhoods”." Criminology 51(2): 287-327.
Hugo, G., Champion, A. and Lattes, A. (2003) Toward a new conceptualization of
settlements for demography. Population and Development Review, 29(2), pp. 277–297.
Available at: https://doi.org/10.1111/j.1728-4457.2003.00277.x.
Jacobs, J. (1984) Cities and the wealth of nations. New York: Random House.
Keil, R. (2021) The density dilemma: there is always too much and too little of it, Urban
Geography, 41(10), pp. 1284–1293. Available at:
https://doi.org/10.1080/02723638.2020.1850025.
Kostof, S. (1991) The City Shaped: Urban Patterns and Meanings Through History. London:
Thames and Hudson.
Kujath, Carlyne L. (2011) Facebook and MySpace: Complement or Substitute for Face-toFace Interaction? Cyberpsychology, Behavior, and Social Networking, 14( 1–2), 75–78.
https://doi.org/10.1089/cyber.2009.0311.
Lieberman, Alicea, and Juliana Schroeder. 2020. ‘Two Social Lives: How Differences between
Online and Offline Interaction Influence Social Outcomes’. Current Opinion in
Psychology, Privacy and Disclosure, Online and in Social Interactions, 31 (February):
16–21. https://doi.org/10.1016/j.copsyc.2019.06.022.
Merrifield, A. (2013) The Urban Question under Planetary Urbanization. International Journal
of Urban and Regional Research, 37(3), pp. 909–922. Available at:
https://doi.org/10.1111/J.1468-2427.2012.01189.X.
Moro, E., Calacci, D., Dong, X., and Pentland, A. (2021) Mobility patterns are associated with
experienced income segregation in large US cities. Nature Communications, 12: pp. 4633.
doi: 10.1038/s41467-021-24899-8
Mukhopadhyay, P., Zerah, M-H, Samanta, G. and Augustin, M. (2016) Understanding
India’s Urban Frontier: What is Behind the Emergence of Census Towns in India? Policy
Research Working Paper No 2923. Washington, DC: The World Bank.
Openshaw, S. and Taylor., P. (1979). A million or so correlation coefficients: three
experiments on the modifiable areal unit problem. Statistical Applications in the Socail
Sciences 21: 127-144.
Parnell, S., & Pieterse, E. (2016). Translational global praxis: rethinking methods and modes
of African urban research. International Journal of Urban and Regional Research, 40(1), 236246.
17
Rohrmeier, K. and Starrs, P.F. (2014) The Paradoxical Black Rock City: All Cities Are Mad.
Geographical Review, 104(2), pp. 153–173. Available at: https://doi.org/10.1111/j.19310846.2014.12015.x.
Samanta, G. (2014). The Politics of Classification and the Complexity of Governance in
Census Towns. Economic and Political Weekly, 49(2), 55–62.
Sayer, A. (1984) Defining the Urban. GeoJournal, 9, pp. 279–285. Available at:
https://doi.org/https://doi.org/10.1007/BF00149040.
Schroeder, J.P. and Pacas, J.D. (2021) Across the Rural–Urban Universe: Two Continuous
Indices of Urbanization for U.S. Census Microdata. Spatial Demography, pp. 1–24.
Available at: https://doi.org/10.1007/s40980-021-00081-y.
Scott, A.J. (2017) The constitution of the city: Economy, society, and urbanization in the capitalist
era. Springer International Publishing. Available at: https://doi.org/10.1007/978-3-31961228-7.
Scott, A.J. (2021) The constitution of the city and the critique of critical urban theory. Urban
Studies, p. 004209802110110. Available at: https://doi.org/10.1177/00420980211011028.
Shin, H.B. (2017) “Rethinking the ’urban’and ‘urbanization,’” in D. Iossifova, C.N.H. Doll,
and A. Gasparatos (eds) Defining the urban: interdisciplinary and professional perspectives.
New York: Routledge, pp. 27–39.
Sinai, T, and Waldfogel, J. (2003) Geography and the Internet: Is the Internet a Substitute or a
Complement for Cities? Journal of Urban Economics, 56(1): 1–24.
https://doi.org/10.1016/j.jue.2004.04.001.
Sorace, C. and Hurst, W. (2016) China’s Phantom Urbanisation and the Pathology of Ghost
Cities. Journal of Contemporary Asia, 46(2), pp. 304–322. Available at:
https://doi.org/10.1080/00472336.2015.1115532.
Spielman, S. and Singleton, A. (2022) A generalized model of activity space. Annals of the
American Association of Geographers 112(8): 2212-2229
Storper, M. and Scott, A.J. (2016) Current debates in urban theory: A critical assessment.
Urban Studies, 53(6), pp. 1114–1136. Available at:
https://doi.org/10.1177/0042098016634002.
Storper, M. and Venables, A.J. (2004) Buzz: face-to-face contact and the urban economy.
Journal of Economic Geography, 4(4), pp. 351–370. Available at:
https://doi.org/10.1093/JNLECG/LBH027.
Stülpnagel, R. v., Brand, D., and Seemann, A.-K. (2019). Your neighbourhood is not a circle,
and you are not its centre. Journal of Environmental Psychology 66: 101439.
Truesdell, Leon E. (1949). The Development of the Urban-Rural Classification in the United
States: 1874 to 1949. Current Population Reports. Population Characteristics.
Washington, DC: US Census Bureau.
United States Census Bureau (USCB) (2022) “Urban Area Criteria for the 2020 Census —
Final Criteria.” United States Federal Register, D.N. 220228–0062
Wirth, L. (1938) Urbanism as a Way of Life. American Journal of Sociology, 44(1), pp. 1–24.
Available at: https://doi.org/10.1086/217913.
Wright, L.E. (2012) Immigration to Tikal, Guatemala: Evidence from stable strontium and
oxygen isotopes. Journal of Anthropological Archaeology, 31(3), pp. 334–352. Available at:
https://doi.org/10.1016/J.JAA.2012.02.001.
Xu, W. (2021) The contingincey of neighbourhood diversity: Variation of social context using
mobile phone application data. Urban Studies, 59(4): pp. 851-869. doi:
18
10.1177/00420980211019637
Yang, Longqi, David Holtz, Sonia Jaffe, Siddharth Suri, Shilpi Sinha, Jeffrey Weston, Connor
Joyce, et al. 2022. ‘The Effects of Remote Work on Collaboration among Information
Workers’. Nature Human Behaviour 6 (1): 43–54. https://doi.org/10.1038/s41562-02101196-4.
Young, I. M. (1990). Justice and the Politics of Difference. Princeton University Press.
19
1
There is a rich literature in urban geography that approaches ‘the urban’ in relation to
theories about the functioning of late capitalism, characterised by authors such as Henri
Lefebvre, David Harvey and Manuel Castells. In this tradition, the ‘urban’ is an expression
of socioeconomic and political relations that transcends strict territorialization. As such, this
genre of urban ontology offers little guidance on the pragmatic question of human
settlement classification.
In some cases, co-presence is not accompanied by co-present social contact, generating a
peculiar form of urbanity. For example, COVID-19 stay-at-home orders heavily constrained
physical copresence in urban centres. We would argue that this does not mean we became
less “urban” during COVID-19. Instead, we collectively experienced a peculiar form of
urbanity characterised by proximity without contact—similar in many ways to modern
prison complexes. Hence, it is important to recognize that it is the demographics, not
permanence or socioeconomic relations, that is the defining feature of an “urban” place.
2
A similar critique underpins the works of Lefebvre, Harvey and Castells, which probe the
entanglements of urban places with late capitalism. As with Wirth, this approach has a
deeply spatial dimension, but obscures the concrete distinctions between different types of
places.
3
4
Why different countries have adopted different definitions is a fascinating and
understudied question. Research to understand the classification logics of national statistical
agencies would likely reveal diverse origins and logics. In some cases, urban criteria might
be legacies of colonial occupation designed to reflect or facilitate ‘development’ activities.
Others may reflect logics of political patronage and control or administrative necessity.
Different cultures may define family, kin and strangers in different ways. Yet in all cases,
the maintenance of meaningful relationships requires time, which is limited. Therefore there
is always an upper bound on the number of meaningful relationships that can be
maintained by every individual.
5
One can debate precisely how long a large group of people must congregate to classify a
location as urban, but there is no conceptual need to restrict our understanding and
definition of urbanity to places of permanent settlement.
6
7
There is a Level 2 Degree of Urbanization classification system that has seven categories,
including urban centres, dense urban clusters, semi-dense urban clusters, suburban cells,
rural clusters, low-density rural cells and very low-density rural cells. This can be used to
classify collections of cells into cities, towns, suburbs and villages. See (Dijkstra et al. 2021)
Indeed, the long chain of post-processing steps (including smoothing and gap-filling) used
in GHS-SMOD to construct boundaries from the process described here indicates how the
arbitrary population size and cell-level density decisions do not capture the local aspects of
the geographic structure of urban places.
8
Absent any specific information on the social networks in communities, the population of
strangers grows as a function of k as shown in Figure 1.
9
The use of alternative experiential “distances,” such as travel time or perceived distance,
also can be easily accommodated and the specific distance metric used to define urbanity is
not intrinsic to this style of local contextual measure of urbanity.
10
Securing the basic provisions for human life and mitigating social conflicts are generally
central.
11
20