Hunger in Conflict: A Quantitative Analysis of the Effects of Armed Conflict on Food Security in West Africa and Afghanistan

Abstract

Half of the world’s undernourished population lives in a country experiencing armed conflict or violence (FAO, 2021). As the main driver of food insecurity, conflict has pushed some 99.1 million people into acute food insecurity in 2020 alone (FSIN, 2021). Additionally, the impacts of COVID-19, climate change, and poverty threaten to exacerbate these already fragile situations, pushing more people towards extreme vulnerability and food crisis (FEWS NET, n.d.). Food insecurity is particularly prevalent in protracted conflicts, where years of violence have worn down government institutions, economies and resources that might otherwise help their populations cope (Peters et al., 2019). Delivering humanitarian assistance is challenging amidst growing needs in increasingly complex environments, and resources are limited. However, response could be improved through a better understanding of how and where violence has the greatest impact on food security, allowing organizations to better target the most vulnerable populations. My research sought to quantify the statistical significance of armed conflict on food security through case studies in West Africa and Afghanistan. I evaluated the duration and intensity of conflict on Integrated Food Security Phase Classification (IPC) levels at both the province and district levels, as well as a temporal analysis through seasonal and timeline regressions.

My study was guided by three main questions: how would IPC levels change with different accumulations of fatalities? How does the duration and intensity of violence impact IPC levels? Which administration levels show the highest correlation between IPC level and violence? I hypothesized that there is a strong correlation between conflict and IPC levels with 12 months of aggregated fatalities, showing the most robust results at an admin 1 level. In addition, I hypothesized that correlations would be stronger in the lean season versus other times of the year. I began my research by producing an attribute table that matched IPC and fatality data to all admin units, which allowed me to analyze data trends and prepare a classification of possible scenarios. Through statistical regression, I quantified the extent to which the duration and intensity of fatalities impacted IPC levels on both province and district levels. Statistical regressions were also performed on seasonal aggregations in addition to a timeline analysis.

The results of the analysis found that the fluctuation in fatality and IPC levels are generally correlated. In Afghanistan this trend was apparent across 24 out of 32 provinces, while in West African countries it was less visible, with three provinces in Nigeria, one in Chad, and none in Mali, Cameroon, or Niger. The strongest correlations were found at the largest admin 1 level with 12 months of aggregated fatalities. However, the results also show that in certain contexts with a prolonged intensity of conflict fatalities, IPC levels and fatalities are correlated and then experience a “tipping point” moment in which IPC levels and fatalities decouple. IPC levels remain high as conflict fatalities decrease. The results of all analyses were transformed into a graduated map that indicates the sensitivity of IPC levels to conflict fatalities. This tool can serve governments, policymakers, and organizations in planning humanitarian food assistance programs, adaptation and mitigation strategies.