Key message 1 – Africa’s urbanisation has not gone hand-in-hand with structural transformation.

Much of the literature on economic growth focuses on structural transformation from agriculture to industry, and by extension, the transition from the countryside to cities (Lewis, 1954; Kuznets and Murphy, 1966). Productivity gains in agriculture are thought to push excess workers out of farming, while emerging urban industries pull labour towards cities.

Cross-country evidence of these patterns is clear (Herrendorf et al., 2014). Take China for example: Structural reforms and the opening up of markets after 1978, brought massive increases in agricultural and manufacturing productivity. In turn, this gave rise to the country’s rapid urban and economic growth. By 2009, the urban population had grown by 700 million people and per capita gross domestic product (GDP) was 13 times larger than it was in 1978 (Angang, 2011).

Typically, manufacturing booms in the largest cities and then diffuses as production techniques become standardised and firms have less to gain from close proximity to workers and markets. Manufacturing relocates to smaller cities, where land and labour are cheaper, while large cities specialise in traded services, where clustering advantages are very strong (Arzaghi & Henderson, 2008; Desmet and Henderson, 2015).

African urbanisation does not conform neatly to these patterns. Many African countries have urbanised without a ‘Green Revolution’, nor the growth of large industrial sectors. As Figure 1 shows, census data suggests there is a high degree of primary sector employment in African cities, regardless of size (Henderson & Kriticos, 2018).1 Even among the largest cities there is a small role for manufacturing and high-value tradable services despite the fact these cities should offer large advantages from economic clustering.

Figure 1: Employment across African cities



  • 1 These estimates are surprising and some contribution may be due to misclassification of census data: Either of locations as “urban” vs “rural”, or of respondents’ employment sector. More research is needed, but these numbers give a preliminary indication.