write summary one paragraph for each article ( 9 articles )

3 4 P o P u l at i o n a n d d e v e l o P m e n t r e v i e w 3 7 ( S u P P l e m e n t ) : 3 4 – 5 4 ( 2 0 1 1 )

The Role of the Demographic Transition in the Process of Urbanization

Tim Dyson

the next observation is, that … there are far more burials than christenings. this is plain.… From this single observation it will follow that london should have decreased in its people, the contrary whereof we see by its daily increase.… it is therefore certain that london is supplied with people from out of the country, whereby not only to repair the over-plus difference of burials above-mentioned, but likewise to increase its inhabitants—John Graunt, Natural and Political Ob- servations made upon the Bills of Mortality. (1662/1964: 35)

as the title suggests, this chapter examines the relationship between the de- mographic transition on the one hand, and urbanization with its attendant urban growth on the other. the chapter is written from the viewpoint that, to their cost, demographers have made rather little out of the demographic transition. after all, this is an incredibly important phenomenon that is af- fecting all of humanity. it has been described, with justification, indeed per- haps with understatement, as “one of the best-documented generalizations in the social sciences” (Kirk 1996: 361). Yet this transition is something that demographers seldom study as a whole. little research specifically tries to link the phenomenon with its many consequences. indeed, demographers seem rather uneasy about wandering too far from their familiar terrain. and one symptom of this is that the urban sector usually enters accounts of the demographic transition only as a determinant of mortality and fertility.1

if one turns to urbanization, again there are major problems with how it is usually approached by social scientists. once more, no one can doubt the great importance of the subject. For example, almost every country in the world is currently urbanizing, and many countries are experiencing unprec- edentedly rapid rates of urban growth. However, these facts are usually taken as the starting point for analysis. inasmuch as research addresses anything further back in the causal chain, it tends to give most attention to the role of (net) rural-to-urban migration in bringing about urbanization. indeed, migration probably receives undue weight in this respect. as others have observed, the causes of urbanization have received relatively little attention (see Preston 1979; woods 2003a).

T i m D y s o n 3 5

in addition, demographers and other social scientists often place too much weight on the interpretation of early experience in their attempts at explanation. this inclination can be especially strong if the early experience relates to their own culture and history. relatedly, there is a tendency to frame explanations in terms of features that eventually turn out to be relatively superficial. Such problems have affected research on both the demographic transition and urbanization. thus, the fact that in european societies in the nineteenth and early twentieth centuries industrialization and modern eco- nomic growth accompanied the demographic transition and urbanization has encouraged the idea that the former sorts of economic processes are the causes of the latter. However, such economic interpretations have faced difficulties in recent decades, because processes like fertility decline and urbanization have been occurring in settings where sustained economic growth and industrial- ization are largely absent.

with this as background, the aim of this chapter is to provide an inte- grated explanation of urbanization (and urban growth) within the context of the demographic transition. to put the matter differently: any account of the demographic transition that fails to include urbanization as one of its major components is seriously incomplete. it is taken for granted here that the process of urbanization has wide-ranging implications for societal development in general (e.g., see dyson 2001). i should note at the outset that the present account of urbanization is far from being original. indeed, as the quotation at the head of the chapter shows, several pieces of the explanation are contained in John Graunt’s statistical analysis of 1662. nevertheless, that account remains unknown to many social scientists and policymakers today.

the chapter has six sections. the first introduces some basic concepts. the second criticizes mainstream economic explanations of both the de- mographic transition and urbanization. the third introduces the stylized sector-specific model of the demographic transition that has been advanced by Jan de vries to help explain the process of urbanization (de vries 1990). the fourth presents empirical examples of how the transition has actually unfolded in urban and rural Sweden and Sri lanka. these examples support and extend the sector-specific demographic account of urbanization. the fifth section considers a corollary of this account—namely that because more recent demographic transitions have involved faster population growth and have occurred over shorter time periods than those that applied in developed countries historically, the contemporary experience of urbanization should also be faster. this hypothesis appears to be upheld. that is, and in contrast to what is often asserted, the general pace of urbanization in contemporary developing countries seems to be relatively fast by historical standards. Fi- nally, the chapter summarizes the conclusions and discusses some of their implications.

3 6 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

Preliminary considerations

the term “urbanization” is used here to refer to an increase in the proportion of a population living in areas that are defined as urban. although different countries use different criteria to classify areas as urban, a key aspect of an urban area is usually that it exceeds a given population size (e.g., 10,000 in- habitants). the full process of urbanization typically involves the movement of a society from being 10 percent urban or less, to being 70 percent urban or more. the term “urban growth” is used here to refer to growth in the number of people who live in urban areas (i.e., the towns). while it is possible for urban growth to happen without urbanization, in the modern world the two processes usually occur together. it is rare, however, for a country’s urban rate of natural increase to greatly exceed the rural rate of increase. indeed, usu- ally the urban rate of natural increase is similar to, or slightly lower than, the rural rate. under these conditions, of course, urbanization requires rural-to- urban migration. this migration reduces the rural rate of population growth and raises the urban rate of growth. therefore it produces faster population growth in the urban sector than in the rural sector—that is, urbanization. ac- cordingly, it is reasonable to say that rural-to-urban migration has been and remains the immediate cause of urbanization in most situations.

Clearly, the fact that different criteria are employed to define areas as urban can complicate matters of comparison—both between different popu- lations and sometimes within the same population over time. For the most part, however, this fact does not constitute a major problem in what follows. another preliminary point is that—even using a fixed set of criteria—in all countries rural areas are reclassified as urban areas from time to time. this reclassification occurs mainly to reflect changes brought about by natural increase and migration—factors that can effectively be regarded as the “real” causes of urban growth. notice, however, that because such reclassification is a periodic administrative process, by its very nature it tends to be lagged on changes brought about by these changes.

Conventional economic accounts

Some comments are required regarding conventional explanations of both the demographic transition and urbanization.

The demographic transition

until fairly recently, explanations of mortality decline and fertility decline within the transition were heavily influenced by the presumed experience of european countries (and their offshoot populations) during the nineteenth and early twentieth centuries.2 in particular, classical demographic transition

T i m D y s o n 3 7

theory, as formulated by Frank notestein (1945, 1953) and others, explained the fertility transition mainly in terms of the emergence of urban-industrial society. this account tended to emphasize the role of economic factors, like rising incomes and the growth of factory employment, in fertility decline. it did not concern itself much with why these societies were becoming in- creasingly urban in the first place—this was seen as being largely due to in- dustrialization. moreover, classical demographic transition theory addressed the urban sector mainly insofar as that sector tended to be associated with relatively low levels of fertility, and tended to experience fertility decline somewhat earlier than rural areas.

in recent decades this classical body of transition theory has had to confront the fact that declines in mortality and fertility have been occurring in poor settings that are neither very urban nor very industrial. Perhaps the most important consequence of this has been the growing recognition— perhaps now a consensus among demographers—that mortality decline is the remote (i.e., underlying) cause of fertility decline.3 this means that the cause of one major demographic process—fertility decline—ultimately lies in another major demographic process—mortality decline. all other consid- erations that may produce fertility declines in different societies are seen as essentially secondary to this.

Urbanization

it is commonly believed that urbanization results chiefly from shifts in em- ployment that occur as a result of economic growth. For instance, addressing the specific causes of urbanization, it has been stated that “the underlying explanation for urbanization involves changing employment opportunities as structural change takes place in the economy” (Jones 2003: 952). this expla- nation emphasizes the shift of the labor force out of agriculture into industry and the service sector that accompanies economic development. in short: economic growth takes off and is sustained in the towns; urban factories of- fer higher wages than rural farms; this attracts people from rural areas; and technological changes resulting from economic growth further reduce work opportunities in agriculture.

again, this classical explanation of urbanization has been heavily af- fected by what happened in europe and north america during the nineteenth and twentieth centuries (e.g., see easterlin 1996; williamson 1988). Countries in these regions experienced industrialization and urbanization at around the same time. their economies changed from being largely agricultural and rural to being largely industrial and urban. there is no doubt that economic growth in urban areas accelerated out-migration from rural areas. and a similar com- bination of processes has occurred in many developing countries—notably in east asia—during recent decades. accordingly, there has been a propensity to

3 8 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

think that what is being observed is cause (i.e., economic growth) and effect (i.e., urbanization).

Perhaps the most important difficulty with this explanation is that in recent decades urbanization has been happening in places where there is little or no economic growth—in particular, sub-Saharan africa. as a result, econo- mists have found themselves addressing the “puzzle that africa urbanized rapidly despite protracted negative [economic] growth” (Fay and opal 1999: 22). other observers of the region note that “rapid urbanization has preceded industrialization; indeed, the african experience seems to imply that it is com- pletely independent of it” (oucho and Gould 1993: 275). Sub-Saharan africa has sometimes seen some movement out of employment in agriculture, but this has occurred mainly because people have physically moved out of rural areas. instead of taking people into a growing industrial sector, migration to the towns has often led them into dire conditions of unemployment and un- deremployment. indeed, accounting for such migration in the face of these conditions has been the subject of particular attention (e.g., todaro 1981). moreover, in recent decades a significant number of countries in other parts of the world have also experienced urbanization during periods of economic stagnation or decline (Fay and opal 1999: 28). the idea that urbanization is often divorced from economic growth also finds echoes in earlier research— for example, work on latin america in the 1920s and 1930s (e.g., davis and Casis 1946; Hoselitz 1957).

A stylized model of urbanization and the demographic transition

ultimately, a population can only experience the process of urbanization as a result of demographic processes.4 in other words, the composition of a popula- tion can only change from being predominantly rural to being predominantly urban through the operation of mortality, fertility, and migration. with rare exceptions, however, migration and the economic conditions that influence it tend to receive most of the attention. this is particularly true in research on urbanization (and urban growth) in the contemporary developing world. there is a widespread tendency to see migration as the principal cause of urban growth and to neglect the often large or even dominant contribution made by urban natural increase. there is also a tendency to view rapid urban growth in isolation from the occurrence of rapid population growth more generally (for more on this, see Preston 1979).

the stylized sector-specific model of urbanization provided by Jan de vries (1990) was proposed on the basis of the historical experience of europe. it was also prompted by the failure of theory on the “mobility transition” to take due account of the demographic transition.5 Key features of the model are found elsewhere—for example, in the work of Bairoch (1988) and wrig-

T i m D y s o n 3 9

ley (1987). But de vries seems to be the first scholar to have drawn things together within the framework of the demographic transition. although he notes that fertility tends to be lower in urban areas than in rural areas, most of the explanation is made in relation to mortality, which probably plays a more important role, and about which more is known. the stylized model is reproduced as Figure 1. what follows is a sketch of the causal chain that it represents.

in pre-transitional circumstances the crude death rate (Cdr) in urban areas is very high (see Figure 1). infectious diseases dominate the causes of death. these diseases generally thrive in towns, where people live at relatively high densities and interact at comparatively high rates. moreover, the urban death rate is not just very high, it is also higher than the urban crude birth rate (CBr). therefore, the urban sector is a demographic “sink”—that is, in the long run its population would not exist without rural-to-urban migra- tion.6 in these circumstances urban growth is limited, and, as Graunt clearly recognized, to the extent that it occurs it is because of migration “out of the country.” However, the high urban death rate means that there is actually a

21

CBR rural

CDR urban

CDR rural

CBR urban

SOURCE: Based on de Vries (1990: 58).

FIGURE 1 A stylized sector-specific model of the demographic transition

Stage of transition

L e v e l

3

4 0 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

restriction (i.e., a ceiling) on how “urban” any population can become. For example, it is thought that the population of the netherlands was about 30 percent urban during the eighteenth century. this was a high level of urban- ization, which cannot have been too far from the ceiling that then applied in the countries of northwestern europe (e.g., see de vries 1974; Keyfitz 1980; wrigley 1987).

to a large degree, the process of mortality decline within the demo- graphic transition happens because of the reduction of deaths from infectious diseases. therefore the urban death rate falls more rapidly than the rural death rate. a crucial point occurs when the urban death rate falls below the urban birth rate. thus stage 2 in Figure 1 begins when urban natural increase becomes positive, that is, the urban population begins to grow partly as a result of its own natural increase. there is no longer a ceiling on the level of urbanization. rural-to-urban migration is initially the main contributor to urban growth. But as the overall level of urbanization in the population rises, so urban natural increase is likely to become the main contributor. it ap- pears that urban natural increase is likely to become the chief cause of urban growth well before the overall population is 50 percent urban (Keyfitz 1980: 149–156). notice that stage 3 in Figure 1 begins when the urban death rate falls below the rural death rate. this raises the theoretical possibility of what de vries terms “autonomous urbanization”—in which the rate of natural in- crease is greater in urban than in rural areas. of course, should this unlikely situation arise, then urbanization would occur even without migration.7

in reality, however, net rural-to-urban migration takes place through- out the demographic transition. initially, such migration is required for the urban sector to exist. then, in the early stages of the transition—when the urban sector is still small—it is the main source of urban growth. migration may also be the main source of urban growth toward the end of the demo- graphic transition, when the urban rate of natural increase is again low (or negative). of course, mortality decline and natural increase in rural areas also generate various socioeconomic pressures—like reductions in the avail- ability of cultivable land per person, and downward pressure on agricultural wages—that lead to an increase in the number of people who migrate out of the rural sector to live in the towns. Furthermore, there are reasons to believe that the net rural out-migration rate may rise as the demographic transition proceeds (Preston 1979).

the causal chain sketched above provides only an outline of what may ac- tually happen. there is ample scope for variation around this basic theme. note too that the work of others has been referred to in order to elaborate the basic account provided by de vries. nevertheless, the role of the urban death rate in initially impeding the process, and then causing the urban natural increase that contributes so greatly to urban growth, is crucial. in addition, particularly during the transition’s earlier phases, the decline of the rural death rate plays an important role by producing a rise in rural-to-urban migration.

T i m D y s o n 4 1

although the data tend to be fragmentary and scattered, many signs from research in economic and demographic history are consistent with this basic demographic account of urbanization. the largest body of evidence relates to europe, where there are strong reasons to believe that the urban sector was a major “sink” in pre-transitional conditions. For example, research on cities like amsterdam and london indicates that the prevailing death rates were extremely high and very much higher than the birth rates. moreover, the conclusion that the urban sector was once a major demographic sink is not altered much by complications introduced by migration between urban and rural areas (e.g., see de vries 1984; wrigley 1987).8

Scattered evidence for populations elsewhere in the world is also con- sistent with the forgoing account—again, especially in relation to urban areas having very high death rates and being sinks in the past (e.g., see Bairoch 1988; woods 2003b). a particularly impressive body of data exists for Bombay (now mumbai) in the decades around 1900. the city’s infant mortality rate was roughly 50 percent (dyson 1997). essentially, the population of about 900,000 was sustained only because of migration. Census data for 1911 show that 80 percent of the city’s residents were born in rural districts (Bairoch 1988: 449).

Sector-specific illustrations of the transition

it is almost impossible to find a full set of published vital rates for the whole of the demographic transition for any country outside of northwestern europe. even within northwestern europe there appears to be only one—Sweden— where for most of the transition death rates and birth rates were published separately for urban and rural areas from an early time. Beyond europe, prob- ably no data set even begins to approach that of Sweden. However, in 1907 P. arunáchalam, the then registrar General of Ceylon (now Sri lanka), stated with considerable justification in the Reports of the Registrar General of Ceylon on Vital Statistics that the country’s vital registration system was “unique in the east, and has few parallels in the west” (see Ceylon 1907, Administration Reports, p. o21). and, fortunately, the published vital registration data for Sri lanka also throw light on the validity of the basic stylized model.

Sweden

For Sweden, table 1 gives average urban (Städer) and rural (landsbygd) vital rates relating to the period 1750–1960. the figures come from two sources. the first is the Statistical Abstract of Sweden (see Sweden, various years). the editions of this publication contain ten-year averages of death and birth rates based on annual vital rates for urban and rural areas for the period 1821– 1960. the annual figures themselves end in the 1960s—mainly because of changes in the definition of urban and rural areas (made necessary, in part,

4 2 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

by the growth of suburbs). therefore the most recent figures in table 1 per- tain to the 1951–60 decade. the second source of the average vital rates is a reconstruction undertaken by Friedlander (1969). this provides the averages for 1750–1820. it is unclear whether these figures are based on registered vital rates for the urban and rural sectors, or whether they result from some form of backward extrapolation. nevertheless, in reconstructing these rates Fried- lander was chiefly concerned with explaining the country’s fertility transition, not its process of urbanization. Furthermore, Bairoch presents average rates of natural increase for the rural areas of Sweden and for Stockholm for the period 1751–1840 that are consistent with Friedlander’s reconstructed rates (see Bairoch 1988: 240–242). more importantly, however, the basic message of the figures in table 1 remains unchanged even if the figures for 1750–1820 are not used.

Table 1 average annual crude vital rates and crude rates of natural increase per 1000 population for urban and rural Sweden, 1750–1960

Year/period CDR (U) CbR (U) CDR (R) CbR (R) CRNI (U) CRNI (R)

1750 36.4 33.2 26.2 36.3 –3.2 10.1 1760 36.7 31.4 26.5 34.6 –5.3 8.1 1770 39.9 30.6 28.1 33.7 –9.3 5.6 1780 38.8 28.9 26.5 32.1 –9.9 5.6 1790 35.9 30.8 24.2 33.9 –5.1 9.7 1800 40.5 28.0 26.8 31.1 –12.5 4.3 1810 38.7 31.0 24.8 34.2 –7.7 9.4 1820 34.0 31.8 21.9 35.2 –2.2 13.3 1826 34.7 31.6 22.4 35.0 –3.1 12.6 1836 33.7 29.1 21.6 31.7 –4.6 10.1 1846 28.8 29.4 19.7 31.3 0.6 11.6 1856 31.2 32.6 20.6 32.8 1.4 12.2 1866 26.2 32.9 19.3 31.2 6.7 11.9 1876 24.1 32.1 17.3 30.2 8.0 12.9 1886 19.7 31.1 16.4 28.7 11.4 12.3 1896 17.4 27.1 16.1 27.2 9.7 11.1 1906 14.9 25.9 14.9 25.7 11.0 10.8 1916 13.4 20.5 14.6 22.7 7.1 8.1 1926 11.2 14.7 12.5 18.8 3.5 6.3 1936 10.8 12.9 12.1 15.3 2.1 3.2 1946 9.5 19.5 11.1 17.7 10.0 6.6 1956 8.8 15.2 10.6 14.2 6.4 3.6

noteS: Crni denotes the crude rate of natural increase, i.e., the birth rate minus the death rate. u denotes urban and r denotes rural. the rates shown for 1826 onward are ten-year averages of figures published in various editions of the Statistical Abstract of Sweden. the rates for 1826 pertain to the years 1821–30 inclusive, and thus the rates for 1956 pertain to 1951–60. the birth and death rates shown for 1750–1820 come from a reconstruction undertaken by Friedlander (1969). their precise derivation is unclear, but for present purposes they are consistent with the figures presented here for 1826 and 1836 in showing urban natural decrease. while Friedlander’s figures provide greater time-depth, they are not required for the argument made in the text.

T i m D y s o n 4 3

with this as background, Figure 2 plots the Swedish vital rates. it shows that until about 1850 the urban death rate was higher than the urban birth rate (see the shaded area indicated by the letter “a”). therefore the urban rate of natural increase was negative (see letter B). indeed, before the 1850s the urban birth rate was lower than the urban death rate by about 6 per thousand—that is, there was appreciable natural decrease. But the rural rate of natural increase remained positive throughout (see C). So it was rural natural increase, through migration, that maintained the urban population and provided the basis for whatever limited urban growth occurred. note too that before about 1850 the urban birth rate was consistently lower than the rural birth rate. therefore lower urban fertility also helped to explain why the urban sector was a sink. the urban death rate behaved as would be expected. it began to fall from about 1800, and it fell much faster than the rural death rate. indeed, not until around 1900 did the urban death rate fall below the rural death rate (see d). after about the 1890s the urban and rural rates of natural increase were approximately equal, and they fluctuated in similar manner (see e).

–20

–10

0

10

20

30

40

50

19501900185018001750

Year

P e r

th o

u sa

n d

p o

p u

la ti

o n

CDR-U

CRNI-R

SOURCE: See Table 1.

FIGURE 2 Crude death and birth rates and crude rates of natural increase for urban and rural Sweden, 1750–1960

A

D

C

B

E

CBR-U

CDR-R

CRNI-U

CBR-R

4 4 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

according to official figures, Sweden’s level of urbanization remained constant at about 10 percent in the decades before 1850. However, as we have seen, positive natural increase was established in the country’s urban sector from about that time. and this and rural-to-urban migration meant that the level of urbanization rose steadily throughout the second half of the nineteenth century—reaching 21 percent in 1900 and 47 percent by 1950 (Sweden 1955).

So the Swedish data reveal a picture that is in close agreement with the stylized model. it is worth noting that before 1950 the urban population had a low sex ratio (m/f) that was usually in the vicinity of 0.86. it seems reason- able to suggest that this may have contributed to the relatively low level of the urban birth rate—inasmuch as adult women may have had difficulty in finding marriage partners (thomas 1941).

Sri Lanka

the case of Sri lanka is more complicated. the Reports of the Registrar General of Ceylon on Vital Statistics (henceforth Reports on Vital Statistics) published in the country’s annual Administration Reports (see Ceylon, various years) provide most of the data used here. the Reports on Vital Statistics for 1891–1907 are the first to provide vital rates for each of the country’s towns. weighted averages of these town rates form the basis of the urban rates used here. then, starting in 1908, annual vital rates were published for the island’s “urban” areas—a series that extends up to the mid-1960s. However, rates for the rural areas were not provided. therefore for 1908 and later years it was necessary to derive the rural rates from those given for the country as a whole (using the corresponding proportion of the total population that was urban). the rates obtained in this way suggest that the urban areas were a “sink” before about 1920. thus the average urban death and birth rates for 1891–1920 were 35.5 and 29.5 per thousand—implying a negative rate of natural increase of –0.6 percent per year. there is frequent discussion of this matter in the early Reports on Vital Statistics.

Because of a major malaria epidemic in 1935, the vital rates published for the country’s urban areas become biased from about the mid-1930s on- ward. the death rate had been falling in the 1920s and the early 1930s. But the epidemic caused it to rise sharply from about 22 deaths to 37 deaths per thousand. this shook the colonial administration. accordingly, efforts were made to increase the hospitalization of birth deliveries (see langford 1996). as a result, the “urban” birth rate rose from about 39 per thousand in 1930–34, to reach 60 per thousand in the years around 1945, and about 70 per thou- sand by 1955. it is clear that the urban birth rate was rising to these unrealisti- cally high levels simply because rural women were increasingly having their births in hospitals and health centers that were located in the towns.

T i m D y s o n 4 5

Fortunately, as early as the 1890s the Reports on Vital Statistics show a re- markably sophisticated appreciation of the factors that can influence statistical measures of mortality and fertility, potential errors in the registration data, and the steps required to improve the data. one aspect of this was the con- cern expressed regarding the degree to which the urban death rate was being influenced by the presence of certain institutions in towns—especially hospi- tals, but also prisons. therefore from 1908 onward the urban deaths of urban residents were distinguished in the published data. as a result, it is possible to calculate the urban death rate of urban residents. this provides a basis—albeit an imperfect one—on which to adjust the total urban death rates for earlier years (i.e., 1891–1907) so that they can be compared with the urban death rates of urban residents from 1908 onward. urban births to urban residents were distinguished from 1927 onward, and similarly this provides the basis for adjusting the urban birth rate for earlier years.9 notice, however, that the resulting adjusted vital rates for urban residents do not involve an upward ad- justment for the deaths and births occurring to urban residents in rural areas, since vital events in rural areas were not differentiated by place of residence. accordingly, the adjusted urban vital rates summarized in table 2 are under- estimates to some degree, while the rural rates are slight overestimates.10

with this as background, Figure 3 plots the adjusted sector-specific vital rates for Sri lanka. essentially, it reveals another version of the same basic

Table 2 average annual crude vital rates and crude rates of natural increase per 1000 population for urban and rural Sri lanka, 1891–1964

Period CDR (U) CbR (U) CDR (R) CbR (R) CRNI (U) CRNI (R)

1891–95 27.1 25.4 28.4 39.7 –1.7 11.3 1896–00 28.1 24.1 26.9 40.1 –4.0 13.2 1901–05 27.8 24.3 26.6 40.3 –3.5 13.7 1906–10 29.8 24.5 31.0 39.3 –5.4 8.3 1911–15 28.1 23.3 30.9 38.9 –4.8 8.0 1916–20 26.4 22.6 30.7 40.5 –3.8 9.8 1921–25 28.4 27.2 27.7 41.0 –1.2 13.3 1926–30 24.5 30.2 25.2 42.0 5.7 16.9 1931–35 22.2 27.6 25.0 38.3 5.4 13.3 1936–40 19.2 27.7 21.7 37.1 8.6 15.4 1941–45 17.5 28.8 20.3 37.9 11.3 17.6 1946–50 13.4 33.8 14.5 39.8 20.4 25.4 1951–55 12.4 37.0 11.0 38.3 24.6 27.3 1956–60 9.8 32.5 9.4 37.3 22.7 27.9 1961–64 8.7 30.8 8.4 35.5 22.1 27.1

noteS: See note 9 and the notes to table 1. the average rates for 1961–64 relate to four (rather than five) years. the full time series are available from the author upon request ([email protected]). as is mentioned in the text, the original source of the data underlying these figures is Reports of the Registrar General of Ceylon on Vital Statistics contained in the annual Administration Reports (see Ceylon, various years).

4 6 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

story. the adjusted rates suggest that until about 1925 the urban death rate was higher than the urban birth rate (see the shaded area indicated by a). therefore the urban rate of natural increase was negative (see B). the ad- justed urban death and birth rates for 1891–1920 average 27.9 and 24.0 per thousand—implying a negative rate of natural increase of about –0.4 percent per year. So, once more, it was rural natural increase that maintained the urban sector and provided the basis for what limited urban growth there was.11 again, the rural rate of natural increase was positive throughout (see C). notice that the urban birth rate was much lower than the rural birth rate. interestingly, the Reports on Vital Statistics for 1916 attribute this to “a far smaller proportion of females in the towns” (see Ceylon 1916, p. l7). in this context it is worth noting that the sex ratio (m/f) of the urban population across the censuses from 1891 to 1946 was high, averaging 1.29 (Ceylon 1951). the adjusted rates suggest that the urban death rate was only slightly higher than the rural death rate during 1896–1905 (see d). However, these urban rates do not reflect the deaths of urban residents that took place in ru- ral areas (deaths that are necessarily included in the rural rates here, biasing them upward). therefore there are compelling reasons to think that in these

–10

0

10

20

30

40

50

Year

P e r

th o

u sa

n d

p o

p u

la ti

o n

197019601950194019301920191019001890

SOURCE: See Table 2.

FIGURE 3 Crude death and birth rates and crude rates of natural increase for urban and rural Sri Lanka, 1891–1964

A

D

C

B

E

CDR-UCRNI-R

CBR-U CDR-R

CRNI-U

CBR-R

T i m D y s o n 4 7

early years the urban death rate exceeded the rural death rate by appreciably more than is shown in Figure 3.12 Finally, from the late 1940s onward the urban and rural rates of natural increase fluctuate in a similar way—rising initially, and then starting to fall. However, the urban rate of natural increase remained lower than the rural rate because the birth rate was lower in the urban areas (see e).

turning to the statistics on urbanization, the Reports on Vital Statistics indicate that in 1891 the country’s 18 towns contained about 11 percent of the island’s population. By 1925 the figure had risen slightly to around 13 percent. the 1963 census put the level of urbanization at 19 percent (Ceylon 1967). therefore the changed mortality conditions in urban areas from about the mid-1920s onward appear to have facilitated a distinct rise in the level of urbanization. moreover, mortality decline in rural areas led to an increase in landlessness and probably a rise in rural-to-urban migra- tion (Sarkar 1957: 208–225). the 2001 census put Sri lanka’s level of ur- banization at just 16 percent (see Sri lanka 2009), but this decline reflects a change in definition. in this context it is noteworthy that when in 1898 the number of towns on the island was increased to 27, the Reports on Vital Statistics noted that several of the new towns were “little more than villages” (see Ceylon 1898, p. F24). there can be no doubt that if the criteria used to define settlements as urban during the period 1898–1963 were applied in 2001, then the country’s level of urbanization would be far higher than 16 percent.

So the Sri lankan data imply a variant of the stylized model. in particu- lar, although the urban death rate was higher than the urban birth rate, it may not have been exceptionally high in itself. on the other hand, it should be recalled that the adjusted urban death rates here are underestimates. it is almost certain that if more adequate data were available for the 1890s and before, they would reveal significantly higher urban death rates. again, how- ever, there are signs that a sex ratio imbalance in the urban population—in this case a shortage of women—may have contributed to the lower birth rate in urban areas.13 and the urban rate of natural increase—which may well lie between the two figures of –0.4 and –0.6 given above—again suggests ap- preciable natural decrease in the urban sector.

The speed of urbanization—past and present

urbanization is both an integral component and an outcome of the demo- graphic transition. one reflection of this is contained in the assertion that the speed of urban growth in contemporary developing countries is unprec- edented by historical standards. thus rapid urban growth in recent decades must be seen in the context of the rapid population growth that has charac- terized more recent examples of the demographic transition (Preston 1979).

4 8 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

of course, both of the main proximate causes of urban growth—namely, natural increase and rural-to-urban migration—are usually strongly related to a country’s overall rate of population growth.

a more problematic assertion is that the speed of urbanization in con- temporary developing countries (i.e., the rate of change in the proportion urban) is not exceptional by historical standards (e.g., see Preston 1979; united nations 1980; Brockerhoff and Brennan 1998; Cohen et al. 2004). this claim is usually made on the basis of estimates of historical and contempo- rary change in the level of urbanization produced within the united nations system—in particular, the historical estimates provided by Grauman (1977) and the contemporary estimates produced biennially by the united nations (e.g., united nations 2008).14 an early example of this assertion, using these sources, is that of Preston who states:

[u]rbanization in developing countries did not proceed with unusual speed in the quarter-century from 1950 to 1975. in this period the percentage urban grew from 16.7 to 28.0 in developing countries. while this is a rapid increase, it is very similar to the one that occurred in more developed countries during the last quarter of the nineteenth century. Between 1875 and 1900, the percent- age urban of countries now more developed grew from 17.2 to 26.1. (Preston 1979: 196)

more recently, Brockerhoff and Brennan have used these same estimates to state that between 1950 and 2000 the level of urbanization in developing countries rose from 17.3 to 40.7 percent; whereas between 1875 and 1925 the level of urbanization in developed countries increased from 17.2 to 39.9 percent (Brockerhoff and Brennan 1998: 78).

of course, the claim that the speed of contemporary urbanization is similar to that experienced by developed countries historically is complicated by matters of definition. in particular, the assumption is made that one can compare historical and contemporary estimates of urbanization, despite dif- ferences in the definition of urban areas and in the procedures employed to reclassify rural areas as urban. moreover, it is important to note that the same assumption is used in what follows below. nevertheless, that difficulty aside, those who make the claim do not support it with an argument as to why such a similarity might exist. Furthermore, if urbanization results from the demographic transition, then one might expect that the speed of urbanization in recent decades would be somewhat faster than applied in the past. this is because more recent transitions appear to be occurring over shorter time scales than generally applied historically.15 moreover, the pace of population growth in contemporary transitions tends to be much more rapid than in the past. therefore, given any threshold size for a settlement to be classified as urban, other things equal we would expect a country to urbanize faster under conditions of faster population growth.

T i m D y s o n 4 9

in fact, the assertion that the speed of urbanization experienced by developing countries is similar to that experienced by developed countries historically depends rather a lot upon the particular comparison that is made. thus the estimates of Grauman (1977) and the united nations (2008) can also be used to support the following statement: between 1920 and 2010 the level of urbanization in developing countries increased from about 8.7 to 45.3 percent; while between 1830 and 1920 the level of urbanization in now-developed countries increased from 8.8 to 37.1 percent. this contrast, relating to a period of 90 years, suggests a speedier pace of contemporary urbanization.

Figure 4 supports a similar conclusion. the bold line shows how the level of urbanization in the countries of the world’s developed regions is esti- mated to have risen since 1800. the un estimates of urbanization in each of the world’s main developing regions during 1950–2010 are then compared with this line—in each case beginning the comparison at the level of urban-

0

10

20

30

40

50

60

70

80

90

Developed regions

20001950190018501800

Latin America

South-east Asia

East Asia

South-central Asia

West Asia and North Africa

Sub-Saharan Africa

NOTE: The bold line relates to urbanization in the more developed regions. Here it acts as a reference standard. Thus although the trends for the individual developing regions all relate to the period 1950– 2010, they have been plotted at the same starting (i.e., 1950) level of urbanization as applied in the more developed regions. Therefore the speed of urbanization can be compared. SOURCES: Grauman (1977); United Nations (2008).

FIGURE 4 Estimates of urbanization, developed regions 1800–2010, and the comparative experience of individual developing regions 1950–2010

Year (for developed-region urbanization)

P e rc

e n

t u

rb a n

5 0 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

ization that corresponds to the level found in the historical experience of the developed regions.

in the world’s most populous developing region, South-central asia, the pace of urbanization seems to be appreciably slower than applied in developed regions in the past (see Figure 4). in this context it is notable that india—by far South-central asia’s most populous country—has been uncommonly slow in reclassifying rural areas as urban. this is at least partly explained by the fact that people living in municipal (urban) areas of india tend to be subject to higher levels of taxation than people living in rural areas—a consideration that leads to some reluctance to be reclassified as municipal (see dyson and visaria 2004: 115). with respect to west asia and north africa, the increase in urbanization during 1950–2010 is comparable to that experienced by de- veloped countries between 1900 and 1960. However, Figure 4 shows that in each of the remaining developing regions the speed of urbanization seems to have been significantly faster than was the case in developed countries his- torically. this is particularly true for sub-Saharan africa and latin america (including the Caribbean). also, east asia and South-east asia experienced an almost identical increase in urbanization during 1950–2010, from around 16 percent to about 48 percent—an increase that was significantly faster than the historical trend experienced in the developed regions.

Discussion

this chapter has argued that urbanization and urban growth are best compre- hended as resulting from the demographic transition. naturally, the expan- sion of towns also requires increased supplies of food and—especially in cold climates—energy from rural areas. But the most important requirement is neither food nor energy, but people.

there are compelling reasons to suppose that urban populations play a major role in stimulating increased agricultural production in rural areas, as the related processes of urban growth and urbanization get underway. relat- edly, while many scholars see the process of urbanization as resulting from sustained economic growth, there is probably at least as much reason to see sustained economic growth as resulting from urbanization. of course, it is impossible to neatly disentangle the independent influence of each process on the other. nevertheless, in trying to account for urbanization social scientists have often gotten the basic direction of causation wrong. thus, as was inti- mated above, shifts in the structure of employment are perhaps better seen as resulting from the demographic processes that bring about urbanization, rather than urbanization being seen as resulting from shifts in the structure of employment as a result of economic growth. in short, mainstream accounts of both the demographic transition and urbanization have often wrongly considered effects to be causes.

T i m D y s o n 5 1

as was stated above, there is growing agreement that mortality decline is the remote cause of fertility decline. of course, mortality decline also brings about population growth. and fertility decline causes population aging. with the exception of the initial process of mortality decline, the cause of each of these other demographic processes essentially lies in another demographic process. the point of this chapter has been to argue that this also applies in relation to the process of urbanization. in each case, economic and other considerations are essentially secondary influences.

the urbanization and urban growth that are occurring in developing countries are first and foremost outcomes of the demographic transition. in general, modern transitions tend to be faster than those that occurred histori- cally. therefore it is unsurprising that the contemporary experience of urban growth—and, it appears, urbanization—tends to be faster too.

the present approach to these two processes also has implications for policy. For example, Preston made the very important point that developing- country governments, which for understandable reasons are often concerned with slowing rapid urban growth, should consider the provision of family plan- ning services as the most palatable means of helping to do so (Preston 1979: 210–211). However, urbanization is a somewhat different matter. as we have seen, it is an inevitable outcome of the demographic transition. moreover, in contrast to rapid urban growth, urbanization is largely a good thing.

Notes

Figures in this chapter are available in color in the electronic edition of the volume.

this chapter was written as part of a research Fellowship funded by the uK eco- nomic and Social research Council (eSrC award reS-063-27-0159). i am grateful to Chris langford for helpful discussion about Sri lanka.

1 to elaborate some of these points: in contrast to demographers and the transition, economists have made a great deal out of modern economic growth; demographic re- search, like research more generally, tends to be overly compartmentalized; there appears to be no book written in english other than this one that has the demographic transition as its title, although, of course, in French there are the major works by adolphe landry (1934) and Jean-Claude Chesnais (notably 1992).

2 the term “offshoot” here is as used by maddison (2007), the main offshoots being the united States, Canada, and australia.

3 the position has a long history (e.g., see davis 1963). But for recent work consistent with this view see, for example, Chesnais (1992); Hirschman (1994); Kirk (1996); mason (1997); Galloway, lee, and Hammel (1998); wilson and airey (1999); Cleland (2001); dyson (2001); livi Bacci (2001); Casterline (2003); reher (2004); ní Bhrolcháin and dyson (2007).

4 naturally this statement does not cover a change brought about by a purely arbitrary reclassification of rural areas as urban.

5 Zelinsky’s (1971) proposal of a “mobil- ity transition” saw this transition as paralleling the demographic transition. However, both of these transitions were attributed to “modern- ization.” there was little attempt to integrate them—for example, as regards how the com- position of internal migration changes from being mainly rural-to-rural to being mainly urban-to-urban.

6 See Sharlin (1978) for the idea that deaths exceeded births in the towns only as a

5 2 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

result of the presence of temporary migrants— among whom it is suggested there was espe- cially high mortality (and low fertility). this raises the possibility that urban populations might not have fallen in the absence of in- migration. However, there is little evidence to support the idea (e.g., see de vries 1984: 182–197). in short, the position that urban areas were demographic sinks has not been overturned.

7 autonomous urbanization is more a theoretical possibility than an established fact because, as i have noted, the urban rate of natural increase is usually similar to or lower than the rural rate; moreover, migration is almost always rural to urban.

8 See also note 6.

9 in the first decade in which the urban death rate of urban residents can be differenti- ated (i.e., 1908–17) it averaged 80 percent of the total urban death rate. in the first decade in which the urban birth rate of urban residents can be differentiated (i.e., 1927–36) it aver- aged 83 percent of the total urban birth rate. therefore, these figures were used to adjust the total urban vital rates for earlier years—the assumption being that these percentages had remained constant. other smaller adjustments were made to the early registration data, but are of little consequence for present purposes.

10 even in 1925 only about 13 percent of the population lived in urban areas. So the transfer back of events occurring to urban residents in rural areas would have a dispro- portionately large upward effect on the urban

rates, and a much smaller downward effect on the rural rates. discussing circumstances in Colombo in the Reports on Vital Statistics P. arunáchalam states that “the density of population in the town, and the presence of hospitals, lunatic asylums, and prisons, tend to raise the death rate of the town, while a counterbalancing influence is exercised by the retirement of the sick from the town to their homes in the country to die” (see Ceylon 1891, p. F14). also notice here his mention of population density.

11 However, immigrant tamil labor brought in from southern india to work on the tea estates also made a contribution.

12 in this context, for the period 1891– 1905 inclusive, the unadjusted urban and rural death rates are 34.6 and 26.5—a difference that probably overstates the true urban/rural differential because of the inclusion in the urban rate of deaths occurring, for example, to rural residents in urban hospitals. See also note 10.

13 in this context P. arunáchalam refers to “the adult females probably being in a small proportion,” the issue being especially acute in relation to the tamil population in urban areas (see Ceylon 1891, p. F14).

14 it should be noted that Cohen et al. (2004: 92–93) also make use of historical data for seven high-income countries and 17 low- and middle-income countries.

15 whether this statement will be sup- ported by future events in sub-Saharan africa remains to be seen.

References

Bairoch, Paul. 1988. Cities and Economic Development. Chicago: university of Chicago Press. Brockerhoff, martin and ellen Brennan. 1998. “the poverty of cities in developing regions,”

Population and Development Review 24(1): 75–114. Casterline, John. 2003. “demographic transition,” in P. demeny and G. mcnicoll (eds.), Ency-

clopedia of Population. new York: macmillan reference. Ceylon. various years. Administration Reports. Colombo: Government Printer. ———. 1951. Census Volume 1, Part II—Statistical Digest. Colombo: Government Printer. ———. 1967. Census of Population, Ceylon 1963. Colombo: Government Printer. Chesnais, Jean-Claude. 1992. The Demographic Transition—Stages, Patterns and Economic Implica-

tions. oxford: Clarendon Press. Cleland, John. 2001. “the effects of improved survival on fertility: a reassessment,” Population

and Development Review 27(Supp.): 60–92.

T i m D y s o n 5 3

Cohen, Barney, martin white, mark montgomery, terrence mcGee, and Yue-man Yeung. 2004. “urban population change: a sketch,” in m. montgomery, r. Stren, B. Cohen, and H. e. reed (eds.), Cities Transformed: Demographic Change and its Implications in the Developing World. london: earthscan Publications.

davis, Kingsley. 1963. “the theory of change and response in modern demographic history,” Population Index 29(4): 345–366.

davis, Kingsley and ana Casis. 1946. “urbanization in latin america, Part 1,” The Milbank Memorial Fund Quarterly 24: 186–207.

de vries, Jan. 1974. The Dutch Rural Economy in the Golden Age, 1500–1700. new Haven: Yale university Press.

———. 1984. European Urbanization, 1500–1800. london: methuen. ———. 1990. “Problems in the measurement, description, and analysis of historical urbaniza-

tion,” in a. van der woude, a. Hayami, and J. de vries (eds.), Urbanization in History. oxford: Clarendon Press.

dyson, tim. 1997. “infant and child mortality in the indian subcontinent, 1881–1947,” in a. Bideau, B. desjardins, and H. P. Brignoli (eds.), Infant and Child Mortality in the Past. oxford: Clarendon Press.

———. 2001. “a partial theory of world development: the neglected role of the demographic transition in the shaping of modern society,” International Journal of Population Geography 7(2): 67–90.

dyson, tim and Pravin visaria. 2004. “migration and urbanization: retrospect and prospects,” in t. dyson, r. H. Cassen and l. visaria (eds.), Twenty-first Century India—Population, Economy, Human Development and the Environment. oxford: oxford university Press.

easterlin, richard a. 1996. Growth Triumphant: The Twenty-first Century in Historical Perspective. ann arbor: the university of michigan Press.

Fay, marianne and Charlotte opal. 1999. “urbanization without growth: a not-so-uncommon phenomenon,” world Bank Policy research working Paper no. 21412, world Bank, washington, dC.

Friedlander, dov. 1969. “demographic responses and population change,” Demography 6(4): 359–381.

Galloway, Patrick r., ronald d. lee, and eugene a. Hammel. 1998. “infant mortality and the fertility transition: macro evidence from europe and new finds from Prussia,” in m. r. montgomery and B. Cohen (eds.), From Death to Birth: Mortality Decline and Reproductive Change. washington dC: national academy Press.

Grauman, John v. 1977. “orders of magnitude of the world’s urban population in history,” Population Bulletin of the United Nations No.8—1976. new York: united nations.

Graunt, John. 1662/1964. “natural and Political observations made upon the Bills of mortal- ity,” Journal of the Institute of Actuaries 90: 4–61.

Hirschman, Charles. 1994. “why fertility changes,” Annual Review of Sociology 20: 203–233. Hoselitz, Bert F. 1957. “urbanization and economic growth in asia,” Economic Development and

Cultural Change 6(1): 42–54. Jones, Gavin. 2003. “urbanization,” in P. demeny and G. mcnicoll (eds.), Encyclopedia of Popula-

tion. new York: macmillan reference. Keyfitz, nathan. 1980. “do cities grow by natural growth or by migration?,” Geographical

Analysis 12(2): 143–156. Kirk, dudley. 1996. “demographic transition theory,” Population Studies 50(3): 361–387. landry, adolphe. 1934. La Révolution Démographique. Paris: Sirey. langford, Christopher m. 1996. “reasons for the decline in mortality in Sri lanka immediately

after the Second world war: a re-examination of the evidence,” Health Transition Review 6(1): 3–23.

livi Bacci, massimo. 2001. A Concise History of World Population. oxford: Blackwell Publishers. maddison, angus. 2007. Contours of the World Economy, 1–2030 AD. oxford: oxford university

Press.

5 4 D e m o g r a p h i c T r a n s i T i o n a n D T h e p r o c e s s o f U r b a n i z a T i o n

mason, Karen o. 1997. “explaining fertility transitions,” Demography 34(4): 443–454. ní Bhrolcháin, máire and tim dyson. 2007. “on causation in demography: issues and illustra-

tions,” Population and Development Review 33(1): 1–36. notestein, Frank w. 1945. “Population: the long view,” in t. w. Schultz (ed.), Food for the World.

Chicago university of Chicago Press. ———. 1953. “economic problems of population change,” 8th International Conference of Agri-

cultural Economists. london: oxford university Press. oucho, John o. and william t. S. Gould. 1993. “internal migration, urbanization, and popula-

tion distribution,” in K. a. Foote, K. H. Hill and l. G. martin (eds.), Demographic Change in Sub-Saharan Africa. washington, dC: national academy Press.

Preston, Samuel H. 1979. “urban growth in developing countries: a demographic reappraisal,” Population and Development Review 5(2): 195–215.

reher, david S. 2004. “the demographic transition revisited as a global process,” Population, Space and Place 10(1): 19–41.

Sarkar, n. K. 1957. The Demography of Ceylon. Colombo: Government Press. Sharlin, allan. 1978. “natural decrease in early modern cities: a reconsideration,” Past and

Present 79: 126–138. Sri lanka. 2009. Statistical Pocketbook. Colombo: department of Census and Statistics. available

at «http://www.statistics.gov.lk/home.asp». accessed december 2009. Sweden. various years. Statistisk årsbok för Sverige (Statistical Abstract of Sweden). Stockholm:

Sveriges officiella Statistik. ———. 1955. Historisk statistic för Sverige (Historical Statistics of Sweden 1, Population, 1720–1950).

Stockholm: Statistka Centralbyrån. todaro, michael P. 1981. Economic Development in the Third World. london: longman. thomas, dorothy S. 1941. Social and Economic Aspects of Swedish Population Movements, 1750–1933.

new York: macmillan. united nations. 1980. Patterns of Urban and Rural Population Growth. new York: united nations,

department of international economic and Social affairs. ———. 2008. World Urbanization Prospects: The 2007 Revision. new York: united nations. williamson, Jeffrey G. 1988. “migration and urbanization,” in H. Chenery and t. n. Srini-

vasan (eds.), Handbook of Development Economics, Volume 1. amsterdam: elsevier Science Publishers.

wilson, Chris and Pauline airey. 1999. “How can a homeostatic perspective enhance demo- graphic transition theory?,” Population Studies 53(2): 117–128.

woods, robert. 2003a. “urbanization in europe and China during the second millennium: a review of urbanism and demography,” International Journal of Population Geography 9(3): 215–227.

———. 2003b. “urban-rural mortality differentials: an unresolved debate,” Population and Development Review 29(1): 29–46.

wrigley, e. a. 1987. “a simple model of london’s importance in changing english society and economy, 1650–1750,” in e. a. wrigley (ed.), People, Cities and Wealth. oxford: Basil Blackwell, pp. 133–156.

Zelinsky, wilbur. 1971. “the hypothesis of the mobility transition,” Geographical Review 61(2): 219–249.