The Urban Transformation of the Developing World Mark R. Montgomery

Sometime in the next 20 to 30 years, developing countries in Asia and Africa are likely to cross a historic threshold, joining Latin America in having a majority of urban residents. The urban demographic transformation is described here, with an emphasis on estimates and forecasts of urban population aggregates. To provide policy-makers with useful scientific guidance in the upcoming urban era, demographic researchers will need to refine their data sets to include spatial factors as well as urban vital rates and to make improvements to forecasting methods currently in use.

By 2030, according to the projections ofthe United Nations (UN) Population Di-vision (1), each of the major regions of the developing world will hold more urban than rural dwellers; by 2050 fully two-thirds of their inhabitants are likely to live in urban areas. The world’s population as a whole is expected to undergo substantial further growth over the pe- riod, almost all of which is expected to take place in the cities and towns of poor countries. The total urban population of these countries was estimated by the UN Population Division to have been 1.97 billion persons in the year 2000, but that total is projected to increase to 3.90 billion by 2030 and further to 5.26 billion by 2050. This will be an enormous change in both relative and absolute terms. The urban demographic transfor- mation influences and is influenced by four allied trends in economic development worldwide: globalization, which binds cities to each other through international networks; the decentralization of governments of poor countries, which is plac- ing greater responsibilities on local and municipal governments (2); evolving international develop- ment strategies to fulfill the Millennium Devel- opment Goals, which explicitly recognize urban as well as rural poverty (3); and the urban impli- cations of global climate change, which is likely to put large coastal city populations at risk from flooding, storm surges, and other extreme weather events (4, 5).

In spite of its centrality to economic develop- ment, the urban transformation of poor countries has somehow largely escaped the attention of the demographic research community. When it has focused on urbanization, the demographic litera- ture has tended to overstate the role being played by very large cities and has underemphasized the importance of small- and medium-sized cities. The literature has also given insufficient weight to urban natural increase versus rural-to-urban migration as a source of city population growth.

There are policy opportunities here that warrant far more attention than has been given to date (6). In turning belatedly to urbanization, demogra- phers have brought one important issue to the forefront: the poor performance of the methods currently being used to forecast city and urban growth in developing countries (2, 7).

Urban and City Definitions Much of what is known about the demography of the urban transition stems from research con- ducted by the UN Population Division, which since the 1970s has been the sole source of inter- nationally comparable city and urban estimates and projections. The main challenge for its ana- lysts is that of heterogeneity: The national defi- nitions that yield these data vary substantially across countries and over time.

Beijing offers one example of the difficulties (8, 9). For the year 2000, the population of “Beijing” was reported by Chinese authorities to be 11.5 million people. But depending on how the city boundary is drawn, the estimate could have been as low as 8.5 million had the boundary encompassed only the administrative units of the

city proper (depicted in darkest shading in the left-hand side of Fig. 1). The official definition also includes the populations of surrounding city districts, which contain more rural than urban residents but are functionally linked to the city proper. Multiple social, economic, administra- tive, and political judgements come into play in the formulation of such city definitions, and it is not obvious that the adoption of any single definition is advisable. Although the conven- tional urban-rural distinction still retains value, a consensus is emerging that future classifica- tion schemes will need to reserve a place for third categories and degrees of urban-ness, as well as the rural and urban ends of the spectrum (10–12).

Although an international database that would allow for consideration of multiple definitions is not yet in hand, a template for this work has been developed by the Global Rural-Urban Mapping Project (GRUMP), which combines detailed ad- ministrative boundary data with urban and rural population counts for all countries and which has supplemented these data with imagery derived from remote sensing and other geographically coded sources (13, 14). A sample of the GRUMP results can be seen in the right-hand side portion of Fig. 1, in which the irregularly shaped areas depicted in light shading indicate where the ur- ban concentrations of population are located in Beijing province as determined by satellite obser- vation of night-time lights (15). In this case, the physical proximity of lights might serve as a proxy for social, economic, and administrative in- teraction. New measures such as these may aid in the development of alternative urban and city definitions (16, 4, 17–19).

The Urban Population Transition The following analysis has its basis in an October 2006 update of the UN Population Division’s

SPECIALSECTION

Department of Economics, Stony Brook University, Stony Brook, NY 11794, USA. Poverty, Gender, and Youth Program, Population Council, New York, NY 10017, USA. E-mail: [email protected]

Huairou Xian

Miyun Xian

Pinggu Xian

Daxing Xian

Shunyi Qu

Tongzhou Qu

Changping Qu

Handian Qu

Shijinshan Qu

Fengtai Qu

Chaoyang Qu

Beijing Shi

Beijing city district boundary

Fangshan Qu

Mentougou Qu

Yanqing Xian

Beijing county boundary

Beijing GRUMP urban extents

Beijing city proper administrative units

Beijing city district administrative units

Beijing county administrative units

Fig. 1. Administrative Districts of Beijing Province, China.

www.sciencemag.org SCIENCE VOL 319 8 FEBRUARY 2008 761

o n

Ju ne

2 1,

2 01

3 w

w w

.s ci

en ce

m ag

.o rg

D ow

nl oa

de d

fr om

cities database (1). Although the units in which city and town populations are recorded vary a good deal and systematic biases have plagued urban and city population forecasts based on these data, there is little disagreement about the broad patterns and trends.

During the period 2000–2024, the world’s total population is projected to grow by 1.76 billion persons, with some 86% of this growth expected to take place in the cities and towns of developing countries (Fig. 2A). These near-term prospects stand in sharp contrast to what was experienced from 1950 to 1974, an era when rural growth still exceeded urban. The projections sug- gest that relatively little additional rural growth will occur in developing countries (an increase of some 190 million rural dwellers in total from 2000 to 2024) and that the UN anticipates that the rural populations of more-developed countries will continue to decline.

Among the major regions of developing coun- tries, Asia now holds the largest number of urban dwellers and will continue to do so (Fig. 2B). By 2025, Africa will have probably overtaken Latin America in terms of urban totals, moving into second place among the regions. (The urban pop- ulation of developing Oceania is also shown, but with only 1.92 million urban residents as of 2000

and 6.47million urban dwellers projected for 2050, the totals for this region are hardly perceptible.)

In the 1950s, 1960s, and well into the 1970s, regional urban growth rates (Fig. 2C) approxi- mated 4% per annum, although declines were already making an appearance in Latin America. Had the growth rates of this early era been sus- tained, the urban populations of the three regions would have doubled roughly every 17 years. By the year 2000, however, urban growth rates had fallen considerably in each of the three major regions. As Fig. 2C indicates, further growth rate declines are forecast for the first few decades of the 21st century, with urban Latin America pro- jected to approach a state of zero growth. Much as with population growth rates overall in develop- ing countries, the urban growth rates in force before 2000 are substantially higher than the rates thatwere seen during comparable historical periods in the West, with the difference being due to lower urban mortality in present-day populations, stub- bornly high urban fertility in some cases, and an built-in momentum in urban growth that stems from the distinctive age and sex structures be- queathed by in-migration of young adults and past population growth (2). Even if the projected downward trends in growth rates come to pass, by 2050 urban growth rates in Africa would

remain about 2% per annum, a rate that would double the urban population of that region in 35 years.

In each of the developing regions, the urban percentage is advancing in a seemingly inexora- ble fashion, and by 2030 urban majorities are projected to emerge in both Asia and Africa. Despite what is often assumed, when compared with the historical experience in Western coun- tries, these decade-to-decade changes in urban percentages—sometimes termed the pace of urbanization—are not especially large (2). The literature exhibits some confusion on this point, often failing to distinguish rates of urban growth, which are rapid by historical standards, from the pace of urbanization, which falls well within the historical bounds.

What has no historical parallel is the emer- gence of hundreds of large cities, especially in Asia and Latin America, which each have several cities above 10million in population. This remark- able feature of the urban transition has attracted a great deal of interest and seems to have fostered the impression that most urban residents in the developing world live in huge urban agglomer- ations. In fact, of all urban residents in cities of 100,000 and above in the developing world, only about 12% live in megacities, i.e., about 1 in 8 of

A

D

B C

P o

p u

la ti

o n

g ro

w th

( b

ill io

n s)

U rb

an p

o p

u la

ti o

n (b

ill io

ns )

U rb

an p

o p

u la

ti o

n g

ro w

th r

at es

2

1.5

1

.5

0

80

70

60

50

40

30

20

10

0

4

3

2

1

0

8

6

4

2

0

150

100

50

0

Total rural growth Total urban growth

1950−1974 1975−1999 2000−2024 1950

Asia Latin America Africa Oceania

Asia Latin America Africa Oceania

1975 2000 2025 2050 1950 1975 2000 2025 2050

Year Year

P er

ce n

ta g

e u

rb an

Africa Latin America Asia Oceania

19 80 19

90 20

10 20

20 20

00 20

30 20

40 20

50 19

80 19

90 20

10 20

20 20

00 20

30 20

40 20

50 19

80 19

90 20

10 20

20 20

00 20

30 20

40 20

50 19

80 19

90 20

10 20

20 20

00 20

30 20

40 20

50

Africa LAC Asia

34

2 1

44

3 4

160

Cities 1-5 million

Cities 5-10 million

Cities 10 million+

MDCs LDCsMDCs LDCs MDCs LDCs

T o

ta l n

u m

b er

o f

ci ti

es

E

17 7

Fig. 2. (A) Urban population growth in more-developed countries (MDCs) and less-developed countries (LDCs), 1950–2024. (B) Total urban popu- lation by region in developing countries. (C) Growth rates of total urban

population by region in developing countries. (D) Increasing percentage urban in developing countries. (E) Number of cities of 1 million residents or more in developing countries in 2000 by region.

8 FEBRUARY 2008 VOL 319 SCIENCE www.sciencemag.org762

Cities

o n

Ju ne

2 1,

2 01

3 w

w w

.s ci

en ce

m ag

.o rg

D ow

nl oa

de d

fr om

urban residents (Fig. 3A). Smaller cities are generally less well provided with basic services than large cities, such as improved sanitation and adequate supplies of drinking water (2). Rates of fertility and infant and child mortality in small cities can be little different from the rates pre- vailing in the countryside. Their municipal gov- ernments seldom possess the range of expertise and managerial talent found in the governments of large cities. Yet in an era of political decen- tralization, these smaller cities are increasingly being required to shoulder substantial burdens in service delivery and take on a larger share of revenue-raising responsibilities (2). Given all this, it is surprising how often small cities have been neglected in policy discussions (20).

The empirical record suggests that various social and spatial feedback mechanisms cause large cities to exhibit declining rates of popula- tion growth, as illustrated by the cases of Jakarta, Seoul, and Bangkok (Fig. 3B). In offering ex- planations, urban economists emphasize how in- creases in city size drive up rents and the many costs of congestion, discouraging prospective migrants and encouraging business relocation. Urban geographers stress the difficulties of lo- cating and measuring the growth of large cities, noting that faster population growth at an urban periphery, whichmay not necessarily be recorded in growth rate statistics, often ac- companies slower growth in the city center.

Another plausible explanation that receives far too little attention is that city growth rates are driven down over time by declines in ur- ban fertility rates. Research by the UN Population Division, based on a sample of countries providing two or more national censuses, al- lows urban population growth rates to be divided into a natural urban growth component (the difference between urban birth and death rates) and a residual that combines net mi- gration with spatial expansion (21).

The details are complicated and the sample of countries small; nevertheless, the results are strikingly at odds with the usual perception of the sources of urban growth. In developing countries, about 60% of the urban growth rate is attributable to natural growth; the remaining 40% is due to migration and spatial expansion. Recently, a very similar rule was established for India over the 4 decades from 1961 to 2001, with urban natural growth again accounting for about 60% of the total [p. 32 of (22)]. Not surprisingly given its low fertility levels, the tight controls that kept migration in check until the 1980s, and the subsequent unleashing of migration, China pre- sents something of an exception. There, the UN’s estimate puts the contribution of natural urban growth at about 40% of the growth rate total.

Many developing-country policy-makers have expressed greater concern about rates of city growth in their countries than about national population growth, and they have not infrequent- ly acted on such concerns with aggressive tactics aiming to expel slum residents and repel rural-to- urban migrants (6). It is therefore surprising how little attention has been paid to a growth-rate pol- icy of a very different character: urban voluntary family-planning programs. Over the past half- century, such programs have compiled an im- pressive record across the developing world in

facilitating fertility declines and reducing un- wanted fertility. Empirical analysis of developing- country city growth and fertility suggests that when national total fertility rates decline by one child, this is associated with a decline of nearly 1 percentage point in city population growth rates for that country (23). Family-planning programs offer an effective and humane alternative to the ineffective and brutalizing measures that have been applied all too often.

Forecasting City Growth The performance of the UN urban and city population forecasts leaves much to be desired: As Table 1 indicates, they have consistently projected growth rates (and thus population sizes) that are too high (2, 7). The mean percentage fore- cast errors are large for the 20-year- and 10-year- ahead forecasts; for example, the 20-year-ahead forecast for Latin America, made in 1980, proved to be 19.8% too highwhen the region’s 2000 urban population was finally counted. The tendency to overproject is not evident in the UN’s forecasts of total population at the national level, and it per- sists despite the insertion of an algorithm in the city forecasting model designed to slow projected growth rates as city size increases. Diagnosing the source of these errors is difficult given that the UN’s method makes no use of fertility or mor- tality rates (which the UN projects in separate exercises) and has not yet incorporated spatially disaggregated data such as shown for Beijing (Fig. 1). Alternative forecasting methods are now being actively explored (7, 23).

If the details remain in doubt while these scientific issues await resolution, at least the broad outlines of future urbanization can be perceived from the UN figures, as can the items in the research agenda that urgently need attention if demographic data and methods are to provide useful scientific guidance. Perhaps the greatest need on the demographic front is to ensure that the censuses regularly fielded by developing countries are analyzed at the level of small geo- graphic units and the results placed in the hands of the local and municipal governments that will

need to make use of such data to ef- fectively plan for the pace and spatial distribution of future growth. Remote- sensing methods can serve as a val- uable supplementary tool, if not in estimating population as such, then in monitoring the spatial spread of city populations in the intercensal periods.

References and Notes 1. United Nations, World Urbanization

Prospects: The 2005 Revision Population Database (Department of Economic and Social Affairs, Population Division, United Nations, New York, 2005).

2. Panel on Urban Population Dynamics, Cities Transformed: Demographic Change and Its Implications in the Developing World, M. R. Montgomery, R. Stren,

A B 10

8

6

4

2

0

12%

24%

16%

36%

12%

100- 500,000

500,000- 1 million 1-5 million

5-10 million

Over 10 million

C it

y p

o p

u la

ti o

n g

ro w

th r

at e

Year

1950 1960 1970 1980 1990 2000

Seoul Bangkok Jakarta

Fig. 3. (A) Distribution of urban population by city size in developing countries in 2000. (B) City growth rates for Seoul, Bangkok, and Jakarta, 1950–2000.

Table 1. Urban population forecast errors for the year 2000 [from (2)].

Mean percentage forecast errors

1980–2000 1990–2000 1995–2000 East Asia and Pacific (EAP) 3.9 26.7 –2.8 EAP excluding China 18.4 9.8 –0.4 Latin America and Caribbean 19.8 5.4 –0.9 Middle East and North Africa 13.3 6.8 8.5 South Asia 27.2 19.7 2.7 Sub-Saharan Africa 21.8 23.4 5.5

Level of development Low 23.1 18.3 3.2 Lower middle 6.9 26.1 –1.3 Lower middle excluding China 25.6 9.9 3.7 Upper middle 12.8 8.9 0.8

www.sciencemag.org SCIENCE VOL 319 8 FEBRUARY 2008 763

SPECIALSECTION

o n

Ju ne

2 1,

2 01

3 w

w w

.s ci

en ce

m ag

.o rg

D ow

nl oa

de d

fr om

B. Cohen, H. Reed, Eds. (National Academies Press, Washington, DC, 2003).

3. United Nations, The Millennium Development Goals Report 2007 (United Nations, New York, 2007).

4. G. McGranahan, D. Balk, B. Anderson, Environ. Urban. 19, 17 (2007).

5. R. Nicholls et al., Ranking port cities with high exposure and vulnerability to climate extremes: Exposure estimates, OECD Environment Working Papers No. 1, Environment Directorate, Paris (2007).

6. UNFPA, State of World Population 2007: Unleashing the Potential of Urban Growth [United Nations Population Fund (UNFPA), New York, 2007].

7. P. Bocquier, Demogr. Res. 12, 197 (2005). 8. K. W. Chan, Y. Hu, China Rev. 3, 49 (2003). 9. K. W. Chan, Eurasian Geogr. Econ. 48, 383 (2007).

10. T. G. McGee, in The Extended Metropolis: Settlement Transition in Asia, N. Ginsburg, B. Koppel, T. G. McGee, Eds. (Univ. of Hawaii Press, Honolulu, HI, 1991), pp. 3–25.

11. T. Champion, G. Hugo, Eds., New Forms of Urbanization: Beyond the Urban–Rural Dichotomy (Ashgate, Aldershot, UK, 2004).

12. T. Champion, “Where do we stand? Lessons from the IUSSP Working Group on Urbanization 2006,” Center for Urban and Regional Development Studies, Newcastle University, Newcastle-upon-Tyne, UK (2006); http://

eprints.ncl.ac.uk/file_store/nclep_411141727771.pdf (accessed 2 January 2008).

13. D. Balk, F. Pozzi, G. Yetman, U. Deichmann, A. Nelson, in Proceedings of the Urban Remote Sensing Conference of the International Society for Photogrammetry and Remote Sensing (International Society for Photogrammetry and Remote Sensing, Tempe, AZ, 2005).

14. Global Rural-Urban Mapping Project (GRUMP), alpha version, Socioeconomic Data and Applications Center (SEDAC), Columbia University, Palisades, NY. The joint effort involves the Center for International Earth Science Information Network (CIESIN), Columbia University; the International Food Policy Research Institute (IPFRI); the World Bank; and the Centro Internacional de Agricultura Tropical (CIAT) (2005). Data available at http://sedac. ciesin.columbia.edu/gpw/.

15. C. Small, Int. J. Remote Sens. 26, 661 (2005). 16. S. Angel, S. C. Sheppard, D. L. Civco, The Dynamics of

Global Urban Expansion (Transport and Urban Development Department, World Bank, Washington, DC, 2005).

17. C. Small, F. Pozzi, C. D. Elvidge, Remote Sens. Environ. 96, 277 (2005).

18. K. C. Seto, M. Fragkias, Landscape Ecol. 20, 871 (2005). 19. A. Schneider, K. C. Seto, D. Webster, Environ. Plann. B

32, 323 (2005). 20. UN-Habitat, Meeting Development Goals in Small Urban

Centres (UN-Habitat and Earthscan, London, 2006).

21. N. Chen, P. Valente, H. Zlotnik, Migration, Urbanization, and Development: New Directions and Issues, R. E. Bilsborrow, Ed. (UNFPA, New York, 1998), pp. 59–88.

22. K. Sivaramakrishnan, A. Kundu, B. Singh, Handbook of Urbanization in India: An Analysis of Trends and Processes (Oxford Univ. Press, New Delhi, 2005).

23. M. R. Montgomery, D. Balk, Global Urbanization in the 21st Century, E. L. Birch, S. M. Wachter, Eds., The City in the 21st Century (Univ. of Pennsylvania Press, Philadelphia, 2008).

24. Thanks are due to H. Zlotnik, Director of the UN Population Division; T. Buettner, assistant director and chief, Population Studies Branch; and G. Heilig, chief of the Division’s Estimates and Projection Section, for making available the October 2006 version of the UN’s cities database. I also thank D. Balk of Baruch College, V. Mara of CIESIN, L. Forouzan of Baruch, and S. Henning of the UN Population Division for preparing Fig. 1, in which they were guided by the work of K. W. Chan of the Department of Geography, University of Washington, Seattle. The work reported here was supported in part by an award from the William and Flora Hewlett Foundation to the Population Council.

10.1126/science.1153012

PERSPECTIVE

Reproducing in Cities Ruth Mace

Reproducing in cities has always been costly, leading to lower fertility (that is, lower birth rates) in urban than in rural areas. Historically, although cities provided job opportunities, initially residents incurred the penalty of higher infant mortality, but as mortality rates fell at the end of the 19th century, European birth rates began to plummet. Fertility decline in Africa only started recently and has been dramatic in some cities. Here it is argued that both historical and evolutionary demographers are interpreting fertility declines across the globe in terms of the relative costs of child rearing, which increase to allow children to outcompete their peers. Now largely free from the fear of early death, postindustrial societies may create an environment that generates runaway parental investment, which will continue to drive fertility ever lower.

Reproducing in cities has been seen as adifficult enterprise ever since cities werecreated. Cities were, and often still are, havens of disease and crime, characterized by costly and crowded housing, transitory commu- nities, and a lack of kin support among inhabi- tants. Indeed, the notion that the “country” is the best place to rear children dates back to the Romans (1). Urban dwellers faced higher rates of infant mortality than their rural compatriots right up to the early 20th century (2), such that most cities were only maintained by a constant influx of migrants. Infanticide and infant aban- donment were used as means of controlling fam- ily size, and even if rescued, the prospects for foundlings were precarious (3). The famous 18th-

century naval captain and philanthropist Thomas Coram, when home from his seafaring duties, was distressed by the number of abandoned babies he observed on the streets of London, which inspired him to set up England’s first foundling hospital in 1741 (4). Unfortunately, similar institutions in other cities could not sustain such levels of philanthropic support in the face of rising numbers of foundlings.

An alternative to such drastic and cruel mea- sures to defray the costs of child rearing is not to have so many children in the first place. The strategy of limiting fertility through techniques such as delaying marriage and sexual abstinence within marriage were not limited to city life; we know they were first used by rural groups, rang- ing from African pastoralists relying on slow breeding livestock such as camels (5) to prein- dustrial German farmers trying to maintain the integrity of their farms by reducing the numbers

of offspring claiming the inheritance (6). The industrial revolution and its resulting urbaniza- tion seem ultimately to have led to a population- wide desire to curb fertility by these techniques. In the late 19th and early 20th centuries, fertility in Britain plummeted in 70 years, from families of six to eight children being common to fam- ilies of more than four children becoming rare. This fertility decline has since become a global phenomenon, albeit with many local differences. Family sizes shrunk first throughout Europe and then in the Americas and Asia, and by the end of the 20th century, this process finally began in earnest in Africa (7). There is a broad associa- tion between declining fertility and declining mortality across countries, although the latter does not necessarily precede the former across individual states or regions, and fertility has continued its downward trend in the West, long after mortality rates were greatly reduced. Here, I have taken an evolutionary perspective to ex- plain some of the mechanisms that might be underpinning this phenomenon.

The fertility decline in Africa is happening fastest in urban areas. In Ethiopia (Fig. 1), a dif- ference of nearly four children between family sizes in the capital city (Addis Ababa) and in rural areas is the highest such difference seen anywhere (8, 9). Although HIV/AIDS can de- press fertility to some extent (10), this cannot fully explain the urban fertility decline, which started before the HIV epidemic took hold. The rural-urban difference is not specific to Ethiopia (even in Europe, where the rural-urban division is somewhat blurred, fertility is still higher in rural than in urban areas). But examining a case where the difference is so dramatic highlights what underlies this trend. Contraceptive services

Department of Anthropology, University College London, Taviton Street, London WC1H 0BW, UK. E-mail: [email protected]

8 FEBRUARY 2008 VOL 319 SCIENCE www.sciencemag.org764

Cities

o n

Ju ne

2 1,

2 01

3 w

w w

.s ci

en ce

m ag

.o rg

D ow

nl oa

de d

fr om