Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Is there a link between economic and urban growth?

Unfortunately the average urbanization rates over a 31-year period provided by UNDP (1993), as reflected in table 3.4, conceal significantly different trends over shorter sub-periods for many countries, even assuming a reasonable degree of accuracy in the data, something that is frequently not the case. Generally, as shown in the previous chapter, the rates were highest during the immediate post-colonial period, coinciding with the economic boom years of the 1960s and early 1970s. Thereafter, the process slowed down somewhat in many countries, before accelerating again. Economic crisis, growing indebtedness, and the impact of structural adjustment during the 1980s and early 1990s have created a more complex and diverse picture. Some support for this contention is derived from the urbanization rates presented in World Bank (1992), which do distinguish the 1965-1980 period from 1980-1990. For 19 of the 43 African countries included, urban growth was faster during the 1980s than during the preceding period, whereas for 21 it was slower. For the remaining 3, it remained virtually constant. The lowest growth rates in Africa cited for the 1980s were 0.4 per cent in Mauritius (down from 2.4 per cent), 2.9 per cent in Tunisia (down from 4.0 per cent), and 3.1 per cent in Egypt (up from 2.7 per cent). In all the other countries, the figures imply that urban growth remained faster than the total population growth rates, which were overwhelmingly between 3 and 3.5 per cent annually. This implication is open to question, as will become clear below.

However, as also stressed by Rakodi and other contributors in this volume, the lack of detailed and reliable data makes it difficult to write with confidence. Many UN and World Bank figures are estimates, modified data, or projections made on various assumptions. One reason for this is the inadequacy of data produced by national statistical offices. This problem is arguably most acute in Nigeria, where all censuses since 1963 have been officially repudiated as untenable (Simon, 1992, p. 171). The most recent, in November 1991, was carefully designed and executed with the aid of a dusk-to-dawn curfew in an effort to be above reproach. However, the total population figure of approximately 88.5 million, being some 20-30 million lower than previous estimates and extrapolations (The Independent, 20 March 1992), has again proved very controversial and contested. It is also widely felt to be inaccurate and the census may yet be cancelled after all (Nigerian Tribune, 23 November 1994).

It appears that some but by no means all of the primate cities have continued to grow very rapidly, especially in poor countries. This is probably on account of a relative lack of attractive alternative destinations for migrants and the prospects of access to facilities and income-earning opportunities, even during recession and structural adjustment, which have reduced traditional rural-urban income disparities (Jamal and Weeks, 1988, 1993). The Tanzanian case illustrates these issues well. Suggestions that the growth of Dar es Salaam declined during the 1978-1988 intercensal period (Barke and Sowden, 1992), have been sharply criticized by Briggs (1993) on the grounds of deficient and incomplete data and misinterpretation of the available data. Whereas both these papers rely on preliminary census data, Potts (1994, pp. 6-7; see also chap. 13) uses the rather different final census figures in arguing that Dar grew by 4.8 per cent per annum from 1978 to 1988, a significant reduction from the 9.7 per cent rate recorded during the previous intercensal period. Nevertheless, even the slower rate is well above the total population growth rate, implying that net in-migration to Dar remained significant. Barke and Sowden (1992) assert that secondary cities continued to grow very rapidly, but Potts cites the final census report, showing that only four grew faster than Dar es Salaam while seven grew more slowly. Evidence from the census and primary fieldwork by Holm (1995) suggest that Tanzania's intermediate towns grew faster than either large cities or small urban centres during the 1980s, although these rates seem to have slowed since the late 1980s as infrastructure and general living conditions there deteriorated steadily. Even here, the balance between the cost of living and income-earning opportunities was unfavourable. The continued urban residence of migrants is therefore explicable in terms of economic diversification and risk-minimizing strategies by "multi-active" households and extended families divided between town and the rural shamba. Access to services unavailable in rural areas remains a vital part of their equation.

Accra, the Ghanaian capital, was reputed to be experiencing significant net out-migration during the worst period of economic hardship in the early to mid-1980s, as people returned to their traditional stool or family lands. Informed Ghanaian sources suggest (pers. comm.) that this trend reversed again once conditions improved under the second Structural Adjustment Programme and PAMSCAD (the much-criticized Programme of Actions to Mitigate the Social Costs of Adjustment). Clear documentary evidence either way is still hard to come by, but Jeffries (1992, pp. 210-213) highlights how dramatically purchasing power and the real value of the legal minimum wage fell between 1974 and 1984 and from 1988 to 1990, with only a very modest recovery over the intervening years. Only a small proportion of the 46,000 civil servants retrenched between 1987 and 1990 took PAMSCAD loans, but of these, Jeffries suggests, a large proportion used them to facilitate a return to farming. Potts (1994, p. 10) reports census data as suggesting that some urban centres have been experiencing net out-migration to rural areas over a considerable period. The 1984 census data reveal a 3.2 per cent annual national urban growth rate for 1970-1984, a marked decline from the 1960-1970 rate of 4.8 per cent and seemingly consistent with deteriorating urban conditions. Curiously, the World Bank (1992) estimates the 1980-1990 urban growth rate at 4.2 per cent annually, a marked increase over its 3.2 per cent figure for 1965-1980. This does not seem likely.

Returning to the Zambian case, Potts (1994, and chap. 13) reports a steady decline in the overall urban growth rate over the three decades, from 8.9 per cent (1963-1969) to 5.8 per cent (1969-1980) to no more than 3.7 per cent (1980-1990) according to the respective censuses. This she attributes, plausibly enough, to the country's economic decline in the wake of the collapse of the world market price of copper, from which Zambia derives over 90 per cent of its export revenues. The Copperbelt towns have experienced the most dramatic decline in their growth rates and experienced net out-migration, in some cases already since the late 1970s. By contrast, World Bank urban growth estimates for Zambia were 7.2 per cent per annum for 1965-1980 and 6.7 per cent annually for 1980-1988, apparently ignoring the evidence.

Although the economic situation in Zimbabwe is arguably less severe than in the countries cited above, the World Bank (1992) estimates overall urban growth in Zimbabwe for 1980-199() to have been 5.9 per cent per annum, virtually unchanged from the 6.0 per cent rate for 1965-1980. The latter is close to the census-based growth rate of 5.8 per cent per annum for 1969-1980. Harare's annual growth rate during the 1980s was recorded as 6.1 per cent in the 1992 census, although this was somewhat lower than expected in view of restrictions on rural-urban migration and the under-enumeration of urban Africans in censuses taken before independence in 1980. It may not, however, last into the 1990s, as several studies, for example Tevera (1995), indicate clearly how serious the impact of public expenditure cut-backs and other measures adopted under Zimbabwe's intensified Structural Adjustment Programme since 1991 has been upon the ability of Harare's urban poor to meet their basic needs.

The foregoing discussion shows clearly how difficult it is to be precise about any demographic and urban trends in Africa. The same is often true with economic data. Seeking clear relationships is therefore no easy task. Everything is contingent. Much depends on the particular data source, the statistical base used, the accuracy and coverage of surveys and censuses, and the method of inter- or extrapolation used for projections. African censuses have - justifiably in the main - acquired a reputation for being notoriously inaccurate on several of these counts (see chap. 2 and above). Whereas populations and growth rates were commonly underenumerated or underestimated during the 1970s, it is plausible that the reverse was the case during the 1980s and in the 1990/91 census round. The World Bank is known to adjust national census data in line with other evidence and possibly its own preconceptions. Conversely, and without wishing to question her contentions, it seems that Potts, in her 1994 paper, placed great reliance on national census figures, just as she pointed to the uncritical use of 1990 World Bank data by Jamal and Weeks (1993).

The wide disparities between different sources are extraordinary. Part of the problem seems to be that data from different years and for somewhat different periods are being juxtaposed. The importance of comparing like with like is underlined by the rapidity with which the World Bank updates some of its data and estimates. For example, Potts's criticisms are based on Bank estimates for 1980-1988 used by Jamal and Weeks (1993) - presumably taken from the 1990 World Development Report, although Potts does not cite the original source. However, as indicated above with respect to Zimbabwe, for example, the Bank's estimates in its 1992 World Development Report cover the whole decade 1980-1990, are substantially lower than its earlier figures, and are close to the 1992 census data. On the other hand, as suggested earlier, the Bank's 1992 figures for Ghana are less plausible.

The 1980s were a period of unprecedented economic hardship for Africa. The combination of sustained depression in world market prices for the continent's principal primary export commodities, deteriorating terms of trade (especially for non-oil producers), and the long-term effects of inappropriate economic policies and expenditures and corruption proved unsustainable. As has been noted in chapter 2, more countries in Africa than anywhere else had adopted IMF/World Bank structural adjustment and economic recovery programmes (with differing degrees of voluntarism and coercion). Government expenditures were slashed, with the result that many social programmes and sectors suffered declining provision. The impact of these cuts was particularly serious for the most vulnerable groups but also for the middle classes, many of whom lost their jobs. At the same time, sub-Saharan Africa's level of indebtedness increased roughly threefold over the decade, making it the world's most indebted region relative to economic size and structure. In 1990, total debt stood at roughly the value of three years' exports of goods and services. By contrast, the position of the key Latin American debtor countries, such as Mexico and Brazil, which precipitated the "debt crisis," improved markedly (Simon, 1995). The majority of African countries suffered a decline in per capita GNP of up to 2 per cent per annum over the decade, but the most serious fall (- 4.6 per cent annually) was recorded in Côte d'Ivoire, the World Bank's erstwhile model of market-oriented economic growth in West Africa (table 3.5). Very few countries recorded positive economic growth over the decade. Mauritius and Botswana head the list at 6.1 and 5.6 per cent per annum, respectively, but it is noteworthy that some of the poorest countries, e.g. Burundi (1.3 per cent), Chad (3.8 per cent), and Burkina Faso (1.2 per cent), also appear on it. Egypt recorded an annual rate of 1.9 per cent.

War, political turmoil, and periodic famine in rural areas have constituted another important source of urban growth in several African countries since the late 1970s, despite economic crisis and hardship. The floods of displaced people into Maputo, Luanda, Addis Ababa, and Monrovia exacerbated already difficult urban conditions in some of the continent's poorest states. Conversely, urban-based insurrections, as in Mogadishu and most recently in Kigali, precipitated large-scale exoduses of urban residents. How temporary or otherwise these flows prove varies in accordance with local conditions and the duration of fighting. In a similar context, the growing toll of the HIV/AIDS pandemic, being concentrated disproportionately among the younger, most economically active - and often best-educated - age cohorts, will have a marked impact upon economic performance and perhaps even on urbanization rates, while the social and health costs and burdens of care will mount rapidly.

One other (admittedly very crude) way to explore the relationship between urbanization rates and economic conditions is to compare the World Bank's (1992) estimates of urbanization rates for 19801990 and the average annual growth rates of GNP in 1980-1991 contained in World Bank (1993) (table 3.5). Not only are the data subject to inaccuracies and anomalies as already discussed, but many GNP figures omit the important "informal" sector and often also peasant production for domestic consumption. These sectors provide employment and forms of income for a high proportion of Africa's population at the best of times; during economic hardship, though, reliance on them increases still further as people engage in risk-spreading and survival strategies (see also chaps. 10 and 13). In recent years it has become more common to include GNP estimates for the subsistence sector, but these figures are often crude and tend to underestimate the actual position (Simon, 1992). The world's three lowest-income countries in terms of GNP per capita (Mozambique, Tanzania, and Ethiopia) all recorded negative economic growth of between - 0.8 per cent and -1.6 per cent per annum during the 1980s, yet their urban growth rates remained extremely high (but see the discussion above regarding data on Tanzania). That of Ethiopia actually increased to 5.3 per cent per annum from 4.9 per cent between 1965 and 1980. Chad (3.8 per cent) and Burundi (1.3 per cent) both experienced positive economic growth rates, but these were still several per cent below their rates of urban growth, which nevertheless had declined modestly since the 1965-1980 period. There is thus considerable diversity, but nowhere did the rate of economic growth outstrip that of urban centres. The most extreme indicators of possible crisis are Niger, Rwanda, Côte d'Ivoire, and Gabon, which experienced economic growth rates of - 4.1, -2.4, - 4.6, and -4.2 per cent per annum simultaneously with urban growth rates of 7.6, 8.0, 4.5, and 6.2 per cent per annum respectively. These four countries span the World Bank's low-income, lower-middle-income, and upper-middle-income categories. Egypt and Tunisia represent the least unfavourable balances, with economic growth and urbanization rates of 1.9 and 1.1 per cent versus 3.1 and 2.9 per cent, respectively (table 3.5). Again, Egypt is classified as a low-income and Tunisia as a lower-middle-income country. It therefore seems apparent that no particular relationship exists between a country's level of GNP per capita (a proxy for formal economic development) and the rate of either recorded economic growth or urban growth rates.

Table 3.5 Urbanization and GNP growth rates, 1980-1990/91


Average annual growth rate (%)

Urban population

GNP per capita




Low-income economies 3.5a - 3.9a
China and India 2.9a - 5.6a
Other low-income economies 4.7a 5.0a 1.0a
1 Mozambique 10.2 10.4 -1.1
2 Tanzania 11.3 10.5 -0.8
3 Ethiopia 4.9 5.3 -1.6
4 Somalia 5.4 5.6 -
6 Chad 8.0 6.5 3.8
9 Malawi 7.4 6.2 0.1
11 Burundi 6.9 5.5 1.3
12 Zaire 4.9 4.8 -
13 Uganda 4.8 4.4 -
14 Madagascar 5.2 6.4 -2.5
15 Sierra Leone 5.2 5.3 -1.6
16 Mali 4.4 3.7 -0.1
17 Nigeria 5.7 6.0 -2.3
18 Niger 7.2 7.6 -4.1
19 Rwanda 7.5 8.0 -2.4
20 Burkina Faso 4.1 5.3 1.2
22 Benin 8.9 5.1 -0.9
25 Kenya 8.1 7.9 0.3
27 Ghana 3.2 4.2 -0.3
28 Central African Republic 4.3 4.8 -1.4
29 Togo 6.6 6.9 - 1.3
30 Zambia 6.6 6.2 -
31 Guinea 4.9 5.7 -
33 Mauritania 10.6 7.5 -
34 Lesotho 7.5 7.0 - 0.5
37 Egypt 2.7 3.1 1.9
40 Liberiab 6.2 6.1 -
42 Sudanb 5.9 3.9 -
Middle-income economies 3.9a 3.4a 0.3a
Lower-middle-income economies 3.7a 3.6a -0.1a
45 Zimbabwec 6.0 5.9 -0.2
46 Senegal 3.3 4.0 0.1
48 Côte d'Ivoire 7.6 4.5 -4.6
52 Morocco 4.3 4.3 1.6b
53 Cameroon 7.6 5.9 -1.0
56 Congo 3.5 4.7 -0.2
63 Tunisia 4.0 2.9 1.1
71 Botswanac 12.6 9.9 5.6
72 Algeria 3.9 4.8 -0.7
74 Mauritius 2.5 0.4 6.1
79 Angolab 6.4 5.8 -
82 Namibiab 4.6 5.3 -1.2
Upper-middle-income economies 4.2a 3.2a 0.6a
86 South Africa 3.2 3.7 0.7
93 Gabon 7.3 6.2 -4.2
100 Libya 9.8 6.3 -
Low- and middle-income economies 3.7a 6.6a 1.0a
Sub-Saharan Africa 5.8a 5.9a - 1.2a
Middle East & North Africa 4.6a 4,4a - 2.4a

Sources: World Bank (1992,1993).

a. Weighted average.
b. Data for years other than those specified.
c. In 1993, Zimbabwe was classified as a low-income country and Botswana as an upper-middle-income country.

Table 3.4 also contains UNDP projections of urban growth rates during the final decade of the twentieth century. In the majority of cases, these are lower than for the 1960-1991 period; in only 18 is the rate expected to rise. The underlying assumptions and basis for projection are unclear, but are likely to be a combination of improved economic growth prospects and increasing rural poverty. The limitations of using past trends as the basis for making projections are now more widely appreciated (Hardoy and Satterthwaite, 1989).

The foregoing discussion indicates that, although there are clearly relationships between urbanization rates and prevailing economic and social conditions, they are complex and neither entirely consistent nor always predictable. Trying to "read off" current or future urbanization trends from economic data is a risky undertaking. For example, the evidence presented in the discussion above on individual countries and cities broadly supports Jamal and Weeks' (1988, 1993) thesis on the "vanishing rural-urban gap" in Africa in certain periods and in certain localities. But the Zambian and Tanzanian data also show that the fortunes of individual urban centres within a single country can vary considerably over the same period. In other words, one cannot automatically assume that falling in-migration and rising out-migration will occur simultaneously or in similar proportions. Holm's (1995) findings in two Tanzanian intermediate towns are significant in this context. Furthermore, economic and political crises need not necessarily reduce the rate of urbanization; there are several examples of the opposite being true, particularly if the immediate problems are concentrated in rural areas. In extreme conditions, such as those afflicting Mozambique during the 1980s, war and economic collapse combined to increase the rate of urban growth considerably.

Meeting urban basic needs

It is extremely difficult to obtain reliable and up-to-date estimates of the extent of basic needs fulfilment in many third world countries. The data on rural-urban disparities in access to basic health facilities, potable water, and adequate sanitation facilities included in the Human Development Report (UNDP, 1993) are thus useful as broad indicators (table 3.6), despite the many gaps and possible questions of accuracy. Access to safe water and sanitation is defined as "reasonable," i.e. within or in close proximity to the dwelling. Access to health services means within one hour's travel on foot or by other means. Unfortunately, there is no time series available to enable direct correlation with economic trends and urban growth rates as discussed in the previous subsection. However, the disparities in the late 1980s (the data are for some time in the 1987-1990 period in each case) are quite marked across Africa, with only a few exceptions. The figures for rural areas are on a par with the poorest countries and/or those with low levels of human development as measured by the Human Development Index.

For the countries with medium human development levels, the urban figures for all three variables are generally over 90 per cent; in rural areas accessibility may be as low as one-third to one-quarter of the urban figure, especially in respect of potable water, for which urban provision is particularly good. In the low human development category, the picture is more patchy and varied. Several countries, surprisingly including very poor ones such as Mali, Niger, and Burundi, apparently manage 100 per cent accessibility to urban water supplies. Notwithstanding the many gaps, especially in respect of health services, the lowest accessibility is commonly for rural sanitation facilities. Rural-urban disparities on this variable may be as high as 1:13 (Niger) but as low as 1:1.25 (Angola). Given the war devastation in Angola, however, the figure can be little more than notional, probably based on the pre-war situation. Egypt is one country in this category with extremely good provision in both urban and rural areas (apart from rural sanitation). These figures are borne out by recently published results of a 1992 national demographic and health survey. Unsurprisingly, provision in the four urban governo-rates of Cairo, Alexandria, Suez, and Port Said was marginally higher than for all urban areas (National Population Council et al., 1993, pp. 21-23). Overall, the data in table 3.6 provide some concrete evidence for the superior access to social services and basic needs in urban areas, undoubtedly one of the contributory causes of sustained rural-urban migration, as discussed above. We cannot, however, adduce anything direct about how such aggregate statistics relate to the preferences and priorities of individual citizens, or about relative provision in different categories of cities and towns. It is the largest centres on which the next section focuses.


Contents - Previous - Next