Rainfall patterns and human settlement in tropical Africa and Asia compared

Kostadis Papaioannou, Ewout Frankema 15 June 2017



There is ample evidence that the evolution of densely settled agricultural civilisations in sub-Saharan Africa has been constrained by adverse ecological and geographical conditions, such as a relative scarcity of navigable rivers and major inland seas, a relatively severe incidence of tropical diseases (Webb 2009, Alsan 2014), or a dominance of light and heterogeneous soils (Djurfeldt et al. 2005). Jared Diamond (1997) has suggested that the vertical orientation of Africa’s continental axis further complicated the diffusion of domesticated animals and plants across the region. Climatological conditions are also challenging (Gallup et al. 1999). Large parts of sub-Saharan Africa are too dry for the cultivation of the more valuable food crops and can barely sustain scattered pastoral communities.

Africa’s biogeography may have had far-reaching implications for the opportunities to adopt modern productivity-enhancing technologies in agriculture, such as hybridised crop varieties. A heterogeneous composition of soils complicates regeneration efforts based on fertilisers. Large ecological diversity and, relatedly, large variety in food production systems in tropical Africa have prevented the use of ‘silver bullet’ inventions, such as the IR8 rice and Norin 10 wheat varieties (Otsuka and Larson 2014). Ecological diversity of tropical Africa thus requires a different ‘green evolution’, a process of tailor-made interventions that generate lower economies of scale, and which will be more costly to effectuate (Conway 1998, Frankema 2014). As Figure 1 illustrates, productivity growth in African agriculture has substantially lagged behind Asia and Latin America, and although biogeographic factors cannot explain all of this divergence, they certainly played an important role.

Figure 1 Index series of gross per capita cereal production, 1961–2015 (1961=100)

Source: FAOSTAT, Production statistics, data retrieved at 10-05-2016; http://faostat3.fao.org/download/Q/*/E

From rainfall levels to rainfall variation

Sub-Saharan Africa is generally dryer than east and southern Asia, with larger arid areas and rainfall patterns that tend to be more erratic. The World Bank has composed drought indicators showing that a large group of tropical African countries experience droughts more frequently, and that there are significant differences in rainfall variability compared with other parts of the tropical world. Le Blanc and Perez (2008) have shown this statistically in a cross-country study using present-day climate data. When low or high rainfall levels reduce the cultivation choice-set of farmers, there are possibilities to adapt through specialisation in crops that are either drought-resistant or can cope with torrential rains in rainforest areas. However, if the amount and annual spread of rainfall is unpredictable, such processes of adaptation may be undermined. The unpredictability of rainfall raises risks of harvest failures and thus affects cultivation choices (e.g. the preference for drought-resistant crops). Indirectly, these conditions affect long-term investments in soil improvement, transportation networks, and commercial infrastructures.

In new research, we take this research a step further by exploring the relationship between rainfall patterns and human settlement using a more fine-grained dataset for a historical era (1920-1940) that precedes large scale global carbon emissions and the major demographic boom in the developing world (Papaioannou and Frankema 2017). We thus avoid a large volume of the noise inherent in studies using contemporary country-level data. Did farmers in tropical Africa face greater climatological insecurity than their counterparts in tropical Asia? And if so, are these reflected in historical patterns of human settlement?

British colonies in the African and Asian tropics

The records of various British colonial administrations in Africa and Asia allow us to create a consistent dataset of monthly rainfall levels, which we need in order to compute degrees of variation from the mean. We make one exception to our focus on British colonies by including data from the Netherlands’ Indies, which are recorded in a largely similar way. In total our sample includes 141 districts across West and East Africa and 80 districts in South and Southeast Asia (see Figure 2). Other colonial records allow us to distil information on population densities at a district level. We adjusted our population data for the possibility of undercounting in colonial population censuses. The colonial records also provide a large set of so-called control variables which we include in our panel regressions in order to better identify the effect of rainfall shocks on the density of human a settlement in a given area.

Figure 2 Tropical African and Asian colonies included in the sample

The intensity and frequency of climate shocks

To measure the degree of rainfall variation at the district level, we distinguish between the intensity of rainfall shocks and their frequency. The intensity indicates how large the deviation from the long-term mean has been. The frequency shows how often such deviations occur when we set the benchmark at one standard deviation. Our results show that, on both measures of variation, farmers in West and East Africa were confronted with greater insecurity (Figures 3a and 3b). Shocks were both more frequent as well as more intense. In pre-modern rural societies high levels of insecurity are often tackled by higher degrees of population mobility. This may have posed a higher constraint on the rise or temporal stability of agricultural civilisations.

Figure 3a Intensity of shocks

Figure 3b Frequency of shocks

Climate shocks and human settlement  

Were districts with higher rainfall variation also less densely populated? To explore our second question, we need to separate the effect that the average level of precipitation has had on conditions for agricultural production and human settlement, from the effect that shocks may have had. As already noted, tropical Africa was generally dryer than tropical Asia, and this difference alone may account for cross-continental differences in human settlement. Figure 4a indeed shows that the drier areas contained fewer people, and that mean rainfall levels in African districts (the blue dots) tended to be lower than in Asian districts (the red triangles). There clearly was and still is a positive correlation between rainfall means and population density.

Figure 4a Rainfall mean and population density

Figure 4b shows that the year-to-year variation in rainfall was larger in tropical Africa, but also, as we hypothesised, that rainfall variation was negatively associated with population density. To investigate whether this simple correlation would hold when including many other variables capturing biogeographic conditions – including mean levels of rainfall – we ran a series of OLS panel regressions. Our main results suggest that a 1% increase in rainfall variability would, on average, have corresponded with a 20% decline in population density. In an uncontrolled regression specification, rainfall variability explains approximately 14.4% of the overall variation in population density. When we include all control variables in the regression the R2 climbs up to 0.47.

Figure 4b Rainfall variation and population density


Our study adds support to the idea that the climatological challenges posed to agricultural development were larger in tropical Africa than in tropical Asia, and that this may be one of the keys to understanding why large parts of tropical Asia have historically been more densely populated than tropical Africa. In so far as higher degrees of climatological variability posed more severe constraints to the adoption of modern, productivity-enhancing farming technologies, this may also partially account for the diverging trajectories of agricultural development in the post-1960 era. However, the implications of our findings go even further. The increasing frequency and intensity of weather shocks observed today, such as heat waves, droughts, floods, and hurricanes, adversely affect conditions of agricultural production and jeopardise efforts to achieve global food security. A rapidly expanding literature shows that climate-induced food shortages pose severe threats to societal cohesion by triggering civil conflict, raising property crime rates, and increasing migration (Miguel et al. 2004, Papaioannou 2017). The effects of climate change are felt harder in Africa than elsewhere. Our study suggests that African farmers also have a longer experience in coping with climate shocks, even though more evidence is needed to explore whether this has been a structural historical condition, or one specific for the interwar era.


Alsan, M (2014), “The effect of the tsetse fly on African development”, The American Economic Review 105(1): 382-410.

Conway, G (1998), The doubly green revolution: Food for all in the twenty-first century, Ithaca, NY: Comstock Pub Associates.

Djurfeldt, G, H Holmén, M Jirström and R Larson (2005), The African food crisis: Lessons from the Asian Green Revolution, Wallingford, UK.

Diamond, J (1997), Guns, germs and steel: The fates of human societies, New York: WW Norton & Company.

Frankema, E (2014), “Africa and the green revolution: A global historical perspective”, NJAS - Wageningen Journal of Life Sciences 70-71: 17-24.

Gallup, J L, J D Sachs and A D Mellinger (1999), “Geography and economic development”, International Regional Science Review 22(2): 179–232.

Le Blanc, D and R Perez (2008), “The relationship between rainfall and human density and its implications for future water stress in Sub-Saharan Africa”, Ecological Economics 66(2): 319-336.

Miguel, E, S Satyanath and E Sergenti (2004), “Economic shocks and civil conflict: An instrumental variables approach”, Journal of Political Economy 112(4): 725-753.

Otsuka, K and D F Larson (2014), An African green revolution, Springer.

Papaioannou, K J (2017), “Hunger makes a thief of any man: Poverty and crime in British Colonial Asia”, European Review of Economic History 21(1): 1-28.

Papaioannou, K and E Frankema (2017), "Rainfall patterns and human settlement in tropical Africa and Asia compared: Did African farmers face greater insecurity?", CEPR Discussion Paper No. 11795.

Webb, J L A (2009), Humanity's burden: A global history of malaria, Cambridge University Press.



Topics:  Development Economic history Environment

Tags:  Africa, Asia, climate, environment, rainfall, civilisation, ecology, population density, Agriculture, development, colonialism

Postdoctoral Research Associate, Department of International Development, LSE

Professor and Chair of Rural and Environmental History, Wageningen University


CEPR Policy Research