Spatial Inefficiencies in Africa’s Trade Network
What is this research about and why did you do it?
Trade costs in Africa are the highest in the world, a fact that is often attributed to the continent not having enough paved roads. But are the roads that do exist at least in the right place to promote beneficial trade? African history offers a variety of potential reasons for why they may not be: infrastructure built for colonial extractive industries or the often arbitrary drawing of national borders might render today’s road network on the continent inadequate to support efficient trading relationships. In this paper, I investigate this neglected second source of high trade costs in Africa.
How did you answer this question?
I first use spatial data from a variety of sources to construct an economic geography of every country in Africa: I use night lights, census data, and routing information from online routing services to measure travel costs between locations. I then use a simplified version of a recent quantitative trade model (Fajgelbaum and Schaal, 2020) to simulate trade flows and derive the hypothetical optimally re-designed network: how would the social planner optimally re-organise roads within a country to support efficient trade flows. I then compare this optimal network to the current one to investigate whether African roads are indeed in the wrong place.
What did you find?
Comparing optimal and current road networks for every country, I first identify South Sudan, Somalia, and Sudan as the countries with the most inefficient networks on the continent. Somalia could gain up to 6% of welfare if it could somehow redesign its road network optimally. I then zoom in further and find that places close to railroad lines built by the colonial powers to this day still have more road infrastructure than they should optimally have, at the expense of other regions. Colonial powers often built infrastructure to support extractive economies such as mining or military conquest. Today’s infrastructure still mirrors these old lines, even though it would have been optimal to invest elsewhere in the country. I also find suggestive evidence that ethnic favouritism in the years since independence has inhibited making these optimal investments.
Results from optimally reshuffling roads in three African countries. In each network graph, every node represents a grid cell centroid location with radius proportional to the size of its local population. Edges are drawn thicker depending on their allotted infrastructure Ii,k. In the optimal networks on the right, the nodes are coloured based on their relative welfare gains and losses, (population gains and losses for the case of mobile labor), with more areas gaining more.
What implications does this have for the study (research and teaching) of wealth concentration or economic inequality?
My results add to our understanding of how regional inequality can persist for decades or even centuries. Massive infrastructure projects furthered inequality more than 100 years ago, and still to this day African nations are stuck in this suboptimal equilibrium. Methodologically, my paper combines quantitative trade modelling with historical and more recent spatial data – an approach that I believe could be used to study related questions across the world.
What are the next steps in your agenda?
In ongoing work, I explore how regional disparities manifest both at the micro level in the case of bus networks in Jakarta (see Kreindler et al, 2023), as well as the macro level in the case of capacity underutilisation in rural areas in Kenya (see Walker et al, 2024). I am excited to continue working on these issues.
Citation and related resources
Graff, T. (2024). Spatial inefficiencies in Africa's trade network. Journal of Development Economics, Vol 171, October 2024.
Kreindler, G. E., Gaduh, A., Graff, T., Hanna, R. and Olken, B. (2023). Optimal Public Transportation Networks: Evidence from the World’s Largest Bus Rapid Transit System in Jakarta. NBER Working Paper 31369, June 2023.
Walker, M. W., Shah, N., Miguel, E., Egger, D., Soliman, F. S. and Graff, T. (2024). Slack and Economic Development. NBER Working Paper 33055
