“Climate Exposed, Debt Decomposed, Reforms Proposed: Climate Disasters and Financing Solutions for Infrastructure in Sub-Saharan Africa”
This working paper was written by Jean Bergero, Louis Carson, Théo Jesover and Nile Kirke, students on the 2025 intake of the MSc in ‘Climate Change and Sustainable Finance’, a programme offered jointly by EDHEC Business School and Mines Paris – PSL.
Under the academic supervision of Dr Edi Assoumou and with the support of Prof. Nadia Maïzi (TTI.5, Centre for Applied Mathematics, Mines Paris – PSL)
Keywords: Sub-Saharan Africa (SSA); Climate Finance; Infrastructure Debt; Cost of Debt; Blended Finance; Climate Risk Pricing; Climate Hazard Risk Index; Satellite Climate Data; Weather Station Data; Behavioral Climate Sentiment; Development Finance Institutions (DFIs); Basel III; CET1 Capital Requirements; Adaptation Pricing.
Abstract:
“Sub-Saharan Africa faces a large and persistent infrastructure financing gap, while climate change is increasing both the need for adaptation investment and the cost of capital. This thesis asks how infrastructure and adaptation finance can be scaled in Sub-Saharan Africa while keeping the cost of debt low enough to avoid debt distress.
To answer this question, the thesis decomposes the cost of debt itself. It brings three climate-related risk pricing channels and a behavioral climate sentiment channel from the sovereign level to the project level, where the association of these factors with Sub-Saharan African infrastructure debt pricing has not previously been analyzed at scale. The empirical analysis uses a cross-sectional dataset of 230 infrastructure projects across 1996-2025, assembled from the African Development Bank and World Bank Project Appraisal Documents, as well as data from the private infrastructure finance information platform IJGlobal. Relevant variables are extracted from more than 1,100 unstructured PDF documents using an LLM-assisted extraction pipeline. The all-in nominal interest rate clustered at financial close is regressed on hazard-decomposed climate exposure measures for flood, storm, and extreme heat events, an NLPderived behavioral sentiment proxy, lender composition, and standard macroeconomic and project-level controls. The climate hazard risk indicators are constructed using climate data informed by satellite observations, combined with local weather station records.”