Climate change and migration in Sub-Saharan Africa

A traveler walks across the dry, dusty short of a lake toward a dock where another person waits with a boat. On the other side of the lake there are green fields and then more dry landscape

Climate change will lead to very significant changes in the global economy. One of the most concerning is population displacement. Countries in Sub-Saharan Africa lie at the center of this issue due to their rapidly growing populations and their dependence on agriculture.

How will changes in the climate, and thus on crop yields, affect the migration patterns in Sub-Saharan Africa?

Can policymakers reduce welfare losses of this process with migration and trade policies?

In BSE Working Paper 1411, “Climate change and migration: the case of Africa,” Bruno Conte uses a quantitative model and high-resolution spatial data to answer these questions.


Building the dataset

To study how the climate links to migration, agriculture-dependence, and trade, Conte builds a novel, high-resolution spatial dataset covering 42 countries in Sub-Saharan Africa (SSA). It consists of granular data on GDP, agricultural suitability and production of multiple crops, crop prices, transportation infrastructure, trade, and migration.

With the new dataset, the author establishes two empirical facts that show how climate change is expected to affect the economy of Sub-Saharan Africa and how trade and migration barriers could hamper the adaptation to these changes.

Fact 1: Heterogeneous expected changes in agricultural yields across crops and space.

Panel A shows the level changes in average potential yields between 2000 and 2080. Panel B shows the standard deviation of the crop-level yield changes within cells.

Panel A shows the level changes in average potential yields between 2000 and 2080. Panel B shows the standard deviation of the crop-level yield changes within cells.

Panel A shows significant expected changes in crop yields over this century (under the “business as usual” climate change scenario posed by the Intergovernmental Panel on Climate Change). Some areas will see very significant drops in their average agricultural productivity, whereas other areas will experience increases. Hence, agricultural production could move elsewhere as a response to climate change.

Moreover, Panel B shows that within-location changes in suitability are different across crops, suggesting crop switching as a potential adaptation margin for affected farmers in these locations.

Fact 2: Crop yields explain many aspects of the SSA economy, like production and migration.

The author also documents a positive correlation between the observed production of crops across SSA locations and their natural yields, which provides evidence of specialization in production. Combined with Fact 1 above, this suggests that the future patterns of production in the continent could be drastically reallocated as a response to climate change. However, this process could be hampered by trade barriers across the continent, which the author also provides evidence for: compared to production, there is a much lower degree of specialization in trade.

Perhaps most importantly, the author also shows evidence which suggests that migration is already responding to climate change in Africa. Specifically, changes in crop suitability between 1975 and 2000 correlate with migration flows within and across countries. This relationship is stronger for nearby location pairs, suggesting that migration costs (e.g., distance) have been limiting the capacity of people to migrate to better locations in the economy.

Building the model

Conte builds a quantitative spatial model that incorporates the main mechanisms and frictions suggested by the two facts. There are several locations and sectors, all but one representing agricultural crops. Locations vary in their relative suitability for these different crops. There is frictional trade and frictional migration.

The model is calibrated to match several features of the SSA economy in the year 2000, such as trade and migration flows, production of different crops, and GDP per capita, among others.

Before using the model to simulate the future, the author conducts a test of its ability to predict observed climate change effects, finding that the model replicates closely the changes in population during the past. Overall, the fact that past changes in the climate allow the model to explain population changes confirms its ability to produce reliable forecasts for the future.

Climate change simulations and policy experiments

Given that the model can match observed population dynamics, it can be used as a laboratory to run experiments that aim at answering the following questions:

  • What will the population distribution in Sub-Saharan Africa look like in 2080 with and without climate change?
  • What will be the welfare effects of it?
  • What could policymakers do about it?
Panel A shows climate migration at the country level and Panel B shows climate migration at the grid cell level

For the first question, the model is simulated for the end of the century under two scenarios: with and without climate change. The resulting differences in population at any given location are therefore “climate migration” since the only difference between the two scenarios is climate change. These are shown in Panels A and B above. Welfare effects are calculated analogously.

At the country level, Panel A shows large climate migration flows from Western Sahelian countries like Mauritania and Senegal to nearby countries and from South African countries to Tanzania and South Africa. Panel B, which presents grid-cell-level results, shows a high degree of within-country heterogeneity. Countries experiencing large migration outflows also experience a high level of internal migration.

In terms of welfare effects, the aggregate results are tiny, but sizable in distributional terms: most countries are worse off, and a few better off.

The author also uses the model to conduct policy-related experiments. For instance, if removing migration frictions in Africa, the aggregate number of climate migrants would rise from around 14 to approximately 84 million people, and the welfare effects would turn from a decrease of 0.1% to an increase of 8.1%. However, this aggregate gain does not sort out the distributional issues, as inequality persists: the welfare gains are driven by a handful of countries where migrants can move into, but those left behind are remarkably worse off.

These results suggest that reducing barriers to migration could help mitigate aggregate welfare losses of climate change, but at the expense of more migration and high inequality.

Panel A shows real income and Panel B shows the alternative welfare measure

Finally, the author shows that trade policy attenuates this tradeoff. Specifically, if combining the reduction in mobility frictions with trade liberalization, the welfare losses from climate change attenuate in aggregate and distributional terms, as shown in Panels A and B above. Importantly, this last experiment uses the European Union as a benchmark.

This suggests that trade and migration policies similar to those applied in the EU could reduce welfare losses in Sub-Saharan Africa without leaving anyone behind.

Conclusions

Climate change will lead to a significant reallocation of population across SSA, both within and across countries. Under the current state of high migration frictions, this will come hand-in-hand with welfare losses.

However, climate change need not lead to bad economic outcomes if rural economies in Sub-Saharan Africa can adapt to it.

If mobility barriers and trade can be reduced, climate change can be a catalyst for migration out of low-productivity rural locations and set off a process of structural change.