Urban Giants: The Costs of Primacy in Sub-Saharan Africa

4th May 2017

Author: Sebastian Kriticos

Africa’s urbanisation is characterised by its tremendous speed and heavy concentration. Today, urban giants like Lagos and Cairo have amassed more than 20 million inhabitants, whilst other African cities such as Kinshasa, Luanda, and Dar es Salaam (all above 5 million) continue to grow at breakneck speed. How these primary cities evolve, and their influence on the rest of the urban hierarchy will have tremendous impacts for Africa’s future development. 

Cities create undeniable advantages. Concentrating activity allows countries to conserve on economic infrastructure and promotes agglomeration economies which flourish in settings where people and industry closely interact. But as economies develop, concentration tends to decline for two reasons. First, it becomes more affordable to spread infrastructure to hinterland regions; and second, urban scale is self-limiting. Beyond a certain point, rising costs-of-living (commuting costs, pollution, disease transmission) make cities less attractive to producers and consumers, who then flock towards secondary and smaller cities.

It seems unsurprising then, that recent research has focused on the dynamics of rapidly growing megacities in developing countries – where the benefits of scale have most notably been counterbalanced by congestion diseconomies (Ades and Glaeser, 1995). Today in Africa, 60 percent of the population live in slums, compared to 34 percent elsewhere. In Dar es Salaam, 28 percent of residents live at least three to a room; in Abidjan, 50 percent (World Bank, 2015, 2016). Housing, infrastructure, and other capital investments, are consistently failing to keep pace with the concentration of people, making Africa’s cities crowded and uncompetitive (Lall et al., 2017).

Economists tend to measure urban concentration using a proxy variable called primacy – the share of total urban population in a country’s largest city (Davis and Henderson, 2003). Figure 1 maps primacy across the world using UN data for 2015. As a whole, Africa appears to be an outlier. Apart from a few exceptions, all countries below the Sahara have primacy rates above 30 percent and several above 50. However, figure 1 also shows that several countries in Latin America and East Asia also have rates of primacy that are as high if not higher than their African counterparts; suggesting primacy may not only be an African phenomenon, but a developing country challenge. 



Figure 1: Primacy in 2015 (share of urban population in the largest city). (Source: UN World Urbanisation Prospects)



Several other factors play a role in determining primacy. Most notable are income per capita and land size. This might explain why countries such as Rwanda are so much darker than France in Figure 1, even though the latter has around 50 million more urban residents. In fact, when studying primacy in a regression framework, we find that once we control for income per capita, land size, and population, as well as regional dummies, there is no significant evidence to suggest that Africa has higher primacy than other world regions.


The Key Challenge

The real issue is that Africa’s population is concentrating while poor - indeed, strikingly poorer than other developing regions with similar urbanization levels. As Figure 2 shows, when the Middle East and North Africa (MENA) became 40 percent urban in 1968, per capita GDP was around $1,800 in constant 2005 US$. When East Asia reached that mark in 1994, per capita income was at $3,600. Yet today as urbanisation in Africa hovers around 40 percent, GDP per capita remains scarcely above $1,000.



Figure 2: Income per capita when urbanisation reached 40 percent. (Source: World Development Indicators)


Low-income implies there are major deficiencies in infrastructure and institutional capacities that are required to accommodate Africa’s larger urban populations. Particularly in the largest cities, this means congestion and its costs will overwhelm the benefits of urban concentration. In fact, a recent paper by Castells-Quintana (2016) does well to reveal this. Using panel data from 1960-2010, Quintana shows that higher primacy will reduce national GDP until a point where at least 50 percent of urban residents have access to basic services such as sanitation and electricity – yet most African countries fall far, far below that threshold.


Primacy and the African city system

We now turn to the question of how primacy affects the entire system of cities. Figure 3 charts the distribution of population across world regions. Several interesting facts on Africa emerge: first, is the relatively large share of population at each tail of the distribution; suggesting low location opportunities for rural migrants outside of the primate city. Second, is Africa’s clear deviation from worldwide trends in its share of population living in cities with more than 1 million inhabitants. In absolute terms, this implies a considerable lack of secondary city development – cities which some consider to be better facilitators of labour mobility, job creation, and the transition from rural to non-rural activities (Ferre et al, 2010).



Figure 3: Urban population distribution by region (2015). (Source: UN World Urbanisation Prospects)


Why might secondary cities be lagging? Well, aside from their internal costs, primate cities tend to monopolise economic and public investment in their respective countries, thus straining the whole city-system. In a mature system, economic activity would spread evenly across space: standardised manufacturing deconcentrates towards secondary and intermediary cities, whilst large metropolitan areas specialise in high value-added industry and services. But in Africa, manufacturing is generally weak and uncompetitive, which diminishes the role for secondary cities.

Many of these cities will rank low on the priorities and budgets of central governments, making service provision and management of infrastructure particularly inadequate. Typically, these cities have large non-tradable based economies, high levels of informal employment, and growth driven by population and consumption rather than productivity. Hence, unless such cities become more competitive, Africa’s urban growth will continue to be absorbed by primate cities in the coming decades, further accentuating the costs of congestion.


Population growth in Africa’s cities 1990-2010

As a final exercise, we motivate this idea by studying the dynamics of urbanisation across a range of African cities. Our goal is to show how population growth varies by city size after controlling for a wide range of relevant geographical features such as distance to coast, elevation, and rainfall. To do so, we use population census data from the Integrated Public Use Microdata Series (IPUMS) for a sample of 34 African countries over the period 1990 to 2010, with typical census intervals between 9-12 years. We harmonize the IPUMS data with annual data on nighttime light emissions at a fine level of spatial disaggregation in order to map our census data to city level statistics. (See Storeygard (2016) for an example of how this is done)

Table 1 shows the results. The dependent variable is annual growth in city population, for a sample of around 800 unique cities. Columns (1)-(3) show that primate cities are growing faster on average than the rest of the urban hierarchy, whilst those in close proximity to primate cities are also experiencing relatively fast population growth. Once we control for population in the base period in columns (4)-(6), we see an even larger coefficient on the primate dummy as well as a negative coefficient on the measure of initial population, thus indicating fast growth in the largest city but a general trend of mean reversion over the urban system as a whole. Column (6) shows that controlling for initial population, tertiary and secondary cities are the lagging, whilst small towns (the base dummy variable) –  which are likely to be absorbing farm populations –  are growing faster. 



Table 1: Regressions of Population growth by city size (1990-2010). (Principal Source: IPUMS)



Addressing Africa’s urbanisation challenges will require a fine balance of tackling the internal and external costs of primacy. Whilst the local impacts of over-concentration are immediate and tangible, few wish to discuss or confront hinterland neglect, particularly if regional competition against the national capital may challenge a centralised government. Nevertheless, policies to promote decentralisation will need to play a critical role in Africa’s development; allowing secondary and intermediary cities to specialise in manufacturing and other tradable industries, whilst helping to attract global investment. Part of this process will occur naturally over the course of the continent’s growth and spatial transformation, but in many areas, targeted reforms and investment are urgently needed to speed-up the process.



Ades, A. and Glaeser, E. (1995) “Trade and Circuses: Explaining Urban Giants.” Quarterly Journal of Economics 110(1), 195–227.

Castells-Quintana, D. (2016) “Malthus living in a slum: Urban concentration, infrastructure and economic growth.” Journal of Urban Economics 92, 31–47.

Davis, J. C. and Henderson, J.V. (2003) “Evidence on the political economy of the urbanization process.” Journal of Urban Economics 53(1), 98–125.

Ferre, C., Ferreira, F., and Lanjou, P. (2010) “Is There a Metropolitan Bias? The Inverse Relationship between Poverty and City Size in Selected Developing Countries.” Policy Research Working Paper 5508. World Bank, Washington, DC.

Lall, S., Henderson, J.V., and Venables, A. (2017) Africa’s Cities: Opening Doors to the World” Washington DC: World Bank.

Storeygard, A. (2016) "Farther on Down the Road: Transport Costs, Trade and Urban Growth in Sub-Saharan Africa" Review of Economic Studies 83 (3): 1263-1295.

World Bank (2015) “Measuring Living Standards within Cities. Households Surveys: Dar es Salaam and Durban.” Washington, DC: World Bank.

World Bank (2016) “Cote d’Ivoire Urbanization Review. Diversified Urbanization.” Washington, DC: World Bank.

Our regression uses WB data for sample of 121 countries with urban population over 1m. The dependent variable is each country’s rate of primacy in 2015; controls include the log of GDP, population and land size in 2005, as well as continent/region fixed effects. Results available on request to author.