Abstract. We propose a composite drought vulnerability indicator (DVI) that reflects different aspects of drought vulnerability evaluated at Pan-African level for four components: the renewable natural capital, the economic capacity, the human and civic resources, and the infrastructure and technology. The selection of variables and weights reflects the assumption that a society with institutional capacity and coordination, as well as with mechanisms for public participation, is less vulnerable to drought; furthermore, we consider that agriculture is only one of the many sectors affected by drought.

The quality and accuracy of a composite indicator depends on the theoretical framework, on the data collection and quality, and on how the different components are aggregated. This kind of approach can lead to some degree of scepticism; to overcome this problem a sensitivity analysis was done in order to measure the degree of uncertainty associated with the construction of the composite indicator. Although the proposed drought vulnerability indicator relies on a number of theoretical assumptions and some degree of subjectivity, the sensitivity analysis showed that it is a robust indicator and hence able of representing the complex processes that lead to drought vulnerability.

According to the DVI computed at country level, the African countries classified with higher relative vulnerability are Somalia, Burundi, Niger, Ethiopia, Mali and Chad. The analysis of the renewable natural capital component at sub-basin level shows that the basins with high to moderate drought vulnerability can be subdivided into the following geographical regions: the Mediterranean coast of Africa; the Sahel region and the Horn of Africa; the Serengeti and the Eastern Miombo woodlands in eastern Africa; the western part of the Zambezi Basin, the southeastern border of the Congo Basin, and the belt of Fynbos in the Western Cape province of South Africa.

The results of the DVI at the country level were compared with drought disaster information from the EM-DAT disaster database. Even if a cause–effect relationship cannot be established between the DVI and the drought disaster database, a good agreement is observed between the drought vulnerability maps and the number of persons affected by droughts.

These results are expected to contribute to the discussion on how to assess drought vulnerability and hopefully contribute to the development of drought early warning systems in Africa.