Forecasting violence in sub-Saharan Africa: What can we learn?

Peace & Security,Sahel Watch12 Jul 2017John O'Loughlin, Frank Witmer, Andrew Linke

The question of how to reduce conflict most effectively is paramount.  Prominent articles in science journals argue that we must stop climate change since it is creating a volatile environment where resource scarcities make farming and pastoralism harder, which, in turn, leads to more competition, sparking more violence. Our research published in the Journal of Peace Research1 suggests that policy makers who rely on this argument may indeed succeed in reducing future violence, but not nearly as much as if they devoted their efforts to improving governance and fair distribution of benefits.

What might the future of violence look like in sub-Saharan Africa? To make informed decisions about the allocation of limited resources, policy makers must understand present conditions by relying on research that analyses factors affecting those conditions, and develop plausible future scenarios that might result if current and modified trends continue into the future.

Understanding the present

Since sub-Saharan Africa is geographically large and diverse in social, political and environmental conditions, there is no guarantee that understanding the drivers of violent conflict in one locality will translate to another locality hundreds of kilometres away. The context of each place (e.g. the ecosystem, cultural practices, institutional and political factors) can mean that violence triggers can vary significantly in their effects. Such local spatial variability does not, however, mean that overarching trends can’t be identified.

Our research takes a synoptic view of violence by analysing violent events over all of sub-Saharan Africa from 1980 through 2012. The events we consider capture political violence such as riots, protests, violence against civilians, militia attacks, and larger battles between rebel and government factions. We count the number of violent events that take place in a month for each 1° x 1° grid (see the maps below) and generate a statistical model to determine which factors best explain the spatial and temporal distribution of violence.

For this model, we consider a wide range of potential drivers of conflict, from socio-demographic factors (population, infant mortality, political rights) to geographic factors (distance to international borders, presence of a capital city), to climate factors (deviation from normal temperature and precipitation). The relationships between these factors and violent events are then quantified by calculating a multiplier for each factor that can be used in the grid-level forecasts. In creating the model, we control for the growing rate of violence reporting in media due to technological developments and the influence of nearby recent violence.

Our models show that factors such as population, political rights, and temperature deviations are all important components contributing to explaining the spatio-temporal distribution of violence. Some statistical models of conflict intentionally omit socio-demographic and geographic factors in order to focus solely on climate factors.2 Our research shows just how important it is to consider documented factors that lead to violence, enabling us to compare the relative importance of multiple influences on future levels of violence.

Future scenarios

To see what the future distribution of violence might look like in sub-Saharan Africa, we evaluate a handful of future scenarios based on the concept of shared socioeconomic pathways.3 Shared socioeconomic pathways are used to characterize alternative global trends in the evolution of society and ecosystems. We operationalize these pathways by formulating plausible future scenarios that vary population growth assumptions, governance trends and climate projections through 2065. For the climate projections, we use climate simulations that use different anthropogenic emissions scenarios or representative concentration pathways (RCPs) with projected low and high CO2 levels.

In the most optimistic scenario, we assume that future population growth rates will be relatively low (2.2 billion in sub-Saharan Africa in 2065), political rights will improve in all countries and temperatures will only increase modestly. In the most pessimistic scenario, future populations will grow rapidly (3.2 billion in 2065), political rights will worsen and temperatures will soar due to climate change driven by high CO2 outputs.

We then take these future scenarios and evaluate them through our statistical model that was calibrated from the observed violent event data. To evaluate which factors are likely to have the most influence, we start by forecasting violence for the decade 2056-2065 according to the most optimistic scenario. Then, we test the individual effects from the pessimistic scenario: poor governance, high fertility, higher temperatures. In this way, we can isolate the effect of each factor. The maps show that failure to improve governance will have the most dire consequences, as measured by the number of violent events by the middle of the century, compared to scenarios with high fertility rates or with higher temperatures.

These maps show the change in forecast violence for the period 2056-2065 using an optimistic future scenario as a baseline forecast. They display the effects of poor governance, high fertility, and higher temperatures. A similar version of these maps was published in the Journal of Peace Research.1 (click to enlarge)

Forecasting the future is an exercise fraught with uncertainty but valuable when used as a tool to explore the outcomes of future scenarios that vary politically, socially and climatologically. To learn more about the mechanisms leading to conflict, our on-going work in Kenya explores the relative effects of cultural, political and environmental factors in multiple sites across that country.4 What policy makers interested in reducing violent conflict in sub-Saharan Africa can take away from our analysis is that improving governance is likely to be much more effective in reducing future violence than mitigating climate change.


  1. Witmer, F., A. Linke, J. O’Loughlin, A. Gettelman and A. Laing (2017) Sub-national violent conflict forecasts for sub-Saharan Africa, 2015-2065, using climate-sensitive models. Journal of Peace Research 54(2):175–192.
  2. Hsiang, S., M. Burke and E. Miguel (2013) Quantifying the influence of climate on human conflict. Science 341(6151): 1235367_0–1235367_14.
  3. O’Neill, B., E. Kriegler, K. Riahi, K. Ebi, S. Hallegatte, T. Carter, R. Mathur and D. van Vuuren (2014) A new scenario framework for climate change research: the concept of shared socioeconomic pathways. Climatic Change 122(3): 387–400.
  4. Linke, A., F. Witmer, J. O’Loughlin, T. McCabe and J. Tir (2017) Drought, local institutional context, and support for violence in Kenya. Forthcoming in Journal of Conflict Resolution.