Forecasting the Fallout: Anticipating the Humanitarian Impact of Armed Conflict

• In Event: What We Can Know about the Future of War and Its Humanitarian Impacts?

Abstract

The ability to forecast conflicts has long been a core task of peace research, and recent advancements in computational approaches are making such effort still more reliable. However, to fully harness the potential of these tools, it is crucial to expand the scope of forecasting beyond the conflict itself and to consider the broader consequences, such as impacts on health, life opportunities, and gendered security issues. This is highly relevant for several reasons. Firstly, conflicts have long-lasting detrimental effects on a range of human development outcomes, and by measuring the impact of conflict solely as lives lost due to violence, we underestimate how destructive and far-reaching violent conflict is. Secondly, while violent conflicts threaten all people exposed, the impacts are not equal across all members of society, as factors such as age, gender, social class, economic status, and ethnicity influence the vulnerability of individuals exposed to violence. Thirdly, factors such as gender, ethnic, and class inequality have been shown to be catalysts for renewed violence, potentially creating a negative feedback loop and acting as part of a conflict trap. In this article, we will demonstrate how we can combine conventional conflict forecasting with the forecasting of derived consequences using ConflictNet, a deep learning architecture designed to learn and extrapolate patterns regarding both the likelihood and magnitude of violent conflicts. Notably, the architecture is also capable of generating outputs for multiple features beyond the conflict itself, enabling us to forecast derived consequences in addition to expected conflict patterns. Specifically, we here forecast infant mortality, child growth failure, and education level disaggregated on age and sex. As such, our approach allows for a more comprehensive assessment of the impacts of conflict, providing valuable insights for researchers and practitioners alike.

Author

• Simon Polichinel von der Maase, Peace Research Institute Oslo