The Nexus of Food, Drought and Conflict: Evidence from a Case-Study on Somalia

This paper aims at disentangling the mutual link between conflict, drought and food security in Somalia. The analysis is conducted using various indicators for food security and on different (national and sub-national) aggregation levels. The evidence is partly based on data from a household-level survey, collected in various regions in Somalia in 2013. In addition, we use geo-spatial regional and district level data, which combines (geo-referenced) drought data, with information on conflict from the joined ACLED-PRIO database, together with other location-specific variables.

Overall, we find an impact of drought on the percentage underweight individuals for pastoral livelihoods on the regional level. At the same time, drought seems to have a small linear increasing effect on the ratio of rural populations in stressed, crisis, and emergency food security situations, while there seems to be no significant effect for urban populations. Based on household panel data, a negative effect of drought on non-food expenditures is found as well as a negative effect of conflict on non-food expenditures, confirming that these households buy less non-food items when confronted with distressing situations. Furthermore, we find an increasing effect of one-seed, intrastate, and internationalized conflict on the percentage underweight individuals on the regional level. In addition, we also find a negative effect of conflict exposure on food expenditures for pastoral (rural) households, in contrast with urban households. This emphasizes the fact that conflict has a more profound effect on the food security of rural households, notwithstanding their functions as food producers. Finally, on the district level, we do not find substantial evidence that drought triggers conflict. In contrast, on the household level we find strong evidence for this, suggesting that conflict analysis at a lower aggregation level does reveal findings that we may not pick up on at a higher aggregation level.