A Network Analysis of Development Finance Institutions and Private Co-Investment in African Agrifood Systems

Abstract

Blended finance policy operates on the assumption that development finance institutions (DFIs) crowd in private capital. However, existing evidence primarily measures the overall volume of co-financing associated with DFI investment rather than identifying whom DFIs actually invest alongside. To address this gap, we provide what is, to our knowledge, the first network-based test of this assumption for African agrifood systems. Using a hand-verified dataset of 131 African agrifood co-investment deals (2015-2025) assembled from six sources, we construct a co-investment network of 347 investors and 1,188 co-investment ties and test whether DFIs occupy bridging positions between public and private capital. DFIs exhibit mean betweenness centrality more than three times that of other investors (p = 0.016), and 62.3% of DFI ties in the network's main component connect outward to non-DFI investors. Crucially, this effect is concentrated in just three institutions: FMO, the International Finance Corporation, and British International Investment. Excluding them eliminates the DFI advantage entirely (p = 0.113). The pattern also varies geographically and sectorally. Bridging is strong in West Africa, but absent in East and Southern Africa, and technology-enabled agrifood segment businesses attract two to three times more co-investors than primary production. These findings demonstrate that crowding in is an institution-specific capability rather than a categorical property of development finance. Consequently, blended-finance mandates and evaluations should focus on the catalytic performance of individual institutions rather than assuming performance across all DFIs.