A new peer-reviewed study in the Proceedings of the National Academy of Sciences quantifies the relationship between major climate oscillations and armed conflict probability at a resolution that corporate risk planners have not previously had access to, and the findings carry direct implications for how organizations assess physical climate risk in exposed supply chain and operational regions.
Researchers at Rice University have produced the first high-resolution dataset linking major climate patterns to armed conflict onset at a local rather than country level, covering more than 500 conflict events from 1950 to 2023. The study, led by statistics doctoral student Tyler Bagwell alongside climate scientist Sylvia Dee and statistician Frederi Viens, was published in the Proceedings of the National Academy of Sciences and focuses on two climate drivers: the El Nino-Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD). The research confirms a statistical relationship between these predictable climate patterns and elevated conflict risk, but with regional specificity that prior studies had been unable to establish.
The core finding is that not all climate impacts carry equal conflict implications. El Nino conditions increase global armed conflict risk relative to La Nina, but the elevated risk is concentrated in regions where El Nino produces drier conditions. In regions where El Nino is associated with wetter conditions, the study found no credible relationship between the climate pattern and conflict probability. That distinction matters for risk localization: a blanket El Nino warning does not translate uniformly to elevated operational or supply chain risk across all exposed regions.
What the Indian Ocean Dipole Finding Adds to the Risk Picture
The study’s treatment of the Indian Ocean Dipole is a significant addition to the climate-conflict literature. Unlike El Nino, where only one phase is associated with higher conflict risk, both positive and negative phases of the IOD were found to increase conflict probability in regions whose climates are strongly coupled to the system. The IOD primarily affects the Horn of Africa and parts of Southeast Asia, two regions with significant agricultural and resource extraction exposure for global supply chains.
The IOD operates on shorter timescales than ENSO and can shift rapidly between phases, creating what the researchers describe as climate whiplash in already vulnerable regions. That rapid cycling between stress conditions may be more disruptive to social and agricultural stability than a longer, more predictable dry period, and the study’s statistical models reflect that finding. For organizations with sourcing, processing, or distribution operations in IOD-coupled regions, the implication is that seasonal climate forecasts for the Indian Ocean basin warrant more systematic attention than they currently receive in most corporate risk frameworks.
Predictability as a Risk Management Tool
Both ENSO and the IOD are forecastable on seasonal to annual timescales, a characteristic the researchers explicitly flag as an opportunity for preparedness rather than simply a description of past patterns. The study was constructed using state-level logistic panel models with random effects across 10,977 country-year observations from 1950 to 2023, with grid cell-level analysis providing the spatial resolution necessary to move beyond national aggregation. The dataset itself required manual geolocating of each conflict event from primary sources including news reports in multiple languages, a process the research team describes as taking up to an hour per case.
The practical application for corporate risk teams is that ENSO and IOD phase forecasts, which American and European meteorological agencies publish months in advance, can serve as leading indicators for elevated instability risk in exposed operating regions. The study does not establish that climate causes conflict directly. It establishes that these climate patterns shift the probability of conflict onset in specific regional contexts, and that those shifts are statistically detectable and geographically localizable. For procurement and supply chain teams doing scenario planning for sourcing regions in sub-Saharan Africa, the Horn of Africa, South and Southeast Asia, and parts of South America, that probability shift is a material input that most current risk models do not incorporate.