Spatiotemporal Dynamics of Conflict Occurrence and Fatalities in Ethiopia: A Bayesian Model and Predictive Insights Using Event-level Data (1997 – 2024)
Yassin Tesfaw Abebe, Abdu Mohammed Seid, Lassi Roininen, Mohammed Seid Ali · 2025
Read Spatiotemporal Dynamics of Conflict Occurrence and Fatalities in Ethiopia: A Bayesian Model and Predictive Insights Using Event-level Data (1997 – 2024) by Yassin Tesfaw Abebe, Abdu Mohammed Seid, Lassi Roininen, Mohammed Seid Ali (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-00867.txt.
Introduction
Yassin Tesfaw Abebe and co-authors model conflict deaths in Ethiopia across nearly three decades, from 1997 to 2024, using event level records from the ACLED project. They treat deaths as two linked outcomes: whether an event kills anyone at all, and how many die when it does.
Their Bayesian model mixes measured factors, such as event type, season, and distance to cities and borders, with a latent spatial and time process that captures clustering. Airstrikes, shelling, and attacks carry the highest risk and counts, and communal and rebel actors account for the most deaths. Fatal events are more likely in summer, border proximity raises intensity, and distance from cities lowers it. The results aim to guide protection planning and resource allocation.
Catalog record
| CREATOR | Yassin Tesfaw Abebe, Abdu Mohammed Seid, Lassi Roininen, Mohammed Seid Ali |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 77 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-00867
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Spatiotemporal Dynamics of Conflict Occurrence and Fatalities in Ethiopia: A Bayesian Model and Predictive Insights Using Event-level Data (1997 – 2024) Yassin Tesfaw Abebe, Abdu Mohammed Seid, Lassi Roininen, Mohammed Seid Ali Abstract This study presents a spatiotemporal, dual Bayesian model designed to examine both the occurrence and the number of conflict fatalities using event-level data applied to over 28 years of conflict data from Ethiopia spanning 1997–2024, sourced from the Armed Conflict Location and Event Data (ACLED) project. Conflict-related fatalities are typically composed of two linked dimensions: the binary occurrence of fatalities and the count of fatalities when they occur. The model includes an additive fixed effect component to model the covariates, along with a random effect component to capture the spatiotemporal influences on both outcomes, while allowing for specific effects for each outcome. Covariates considered in the model include event types and season as categorical variables, proximity to cities and borders as nonlinear effects, and population as an offset term in the count model. A latent spatiotemporal process is used to account for common spatial and temporal influences on both outcomes. The spatial structure is modeled using a Matérn field prior, and inference is carried out through Integrated Nested Laplace Approximation (INLA). The results demonstrate significant spatial clustering and temporal fluctuations in fatality risks, reinforcing the value of incorporating both dimensions for a more profound understanding and better prediction of the dynamics of conflict-related violence. For event types, findings showed that airstrikes, shelling, and attacks had the highest likelihood of fatality occurrence and had the highest impact on the likelihood of multiple fatality counts. Foreign, communal, and rebel actors derive the highest conflict fatalities, while protesters and rioters show markedly lower fatality risks, reflecting the distinct impact of actor types. The finding also revealed that there is a higher likelihood of multiple fatalities in the summer season. The study also conclude that proximity to international borders is a primary driver of high-intensity violence, while remoteness from urban centers significantly increases the probability of lower-intensity fatal events. These findings provide new insight into the dynamics of conflict violence and offer practical value for planning, policy, and resource allocation to better protect vulnerable communities. Keywords: Spatiotemporal Analysis, Conflict Fatalities, Bayesian Modeling, INLA, SPDE Approach, ACLED, Ethiopia 1 Introduction Ethiopia, located at the strategic crossroads of the Horn of Africa, has faced recurrent waves of violent conflict rooted in a complex interplay of historical, political, and socioeconomic factors. The country’s modern political and geopolitical history has included transitions from imperial rule to military dictatorship and, more recently, to an ethnically based federal system. Each of these shifts has generated new forms of contention and realigned social and political fault lines [ 25 , 18 ] . Over time, struggles for control over land and resources, competition among political elites, and the politicization of ethnic identity, along with various forms of foreign interventions because of the country’s geopolitical vulnerabilities following its strategic hydropolitical roles over the Nile and the Red Sea, have interacted to produce both localized and widespread outbreaks of violence. Persistent divisions of various sections of the society, poor democratic culture, and politicization of ethnicity have further amplified conflict, resulting in patterns that are highly dynamic [ 45 , 48 ] . The above socio-economic, political, and strategic geopolitical factors embedded in the country’s political history, along with limitations of the current administrations of the Prosperity Party (PP), have made political stability nearly impossible in Ethiopia even after the post-2018 political reform but have instead been followed by increasing intensity and types of conflicts across different parts of the country. The Tigray war (November 2020 – November 2022) emerged as one of the world’s deadliest contemporary civil wars; estimates indicate hundreds of thousands of fatalities, including extensive civilian casualties [ 33 ] . The humanitarian consequences were severe; more than 2.6 million people were internally displaced, and hundreds of thousands more fled across borders. Critical infrastructures, including hospitals, schools, and utilities, were destroyed [ 24 ] . The violence subsequently spread to the Afar and Amhara region, resulting in millions more displaced, destruction of private properties and civilian infrastructures, serious violation of humanitarian law including repeated targeting of civilian populations, various forms of sexual violence, and mass killings of innocent civilians as an act of ethnic violence against the Amhara and Afar communities [ 50 ] . Almost immediately after the cessation of hostilities in Tigray following the Pretoria Accord in November 2022, a new conflict front erupted in the Amhara National Regional State, where the Fano militia, formerly allied with the federal government, launched a widespread insurgency in 2024. Heavy fighting with the Ethiopian National Defense Force has led to significant casualties on both sides, while civilians have suffered from extrajudicial killings and aerial drone strikes that have struck residential areas, schools, and health centers. This renewed violence has displaced more than two million people in the Amhara region alone, overwhelming food systems, collapsing healthcare, and disrupting education for millions of children [ 50 , 53 ] . Simultaneously, persistent communal violence in the Oromia, Somali, Afar, and Benishangul-Gumuz regions has displaced millions more, contributing to an ongoing, nationwide humanitarian emergency [ 5 ] . These widespread crises are compounded by the weakness of state institutions and state fragility, corruption, ethnic politicization, and interference by neighboring states and non-state actors, all of which create opportunities for armed groups to aggravate the fragility of the Ethiopian state [ 34 ] . The cumulative impact of these concurrent conflicts has resulted in a humanitarian crisis of a wider scale. By 2024, over 4 million Ethiopians required urgent humanitarian assistance as attacks on health and educational facilities became commonplace, crippling access to basic government services. Agricultural production, the mainstay of most livelihoods, has been severely disrupted, raising the specter of famine. The educational crisis is acute: 4.4 million children are out of school in Amhara and Oromia alone [ 21 ] . Moreover, violent conflict has a profound impact on food security, health, and economic stability. A study Abay et al. 2023 found that seven months into the conflict, the probability of moderate to severe food insecurity increased by 37 percent, with each additional battle exposure leading to a 1% rise in this probability. Another study in conflict and health reported that 57% of households in war-torn Tigray experienced food insecurity, with 30.47% mildly, 27.21% moderately, and 21.86% severely food-insecure [ 24 ] . The study Arage et al. 2023 reveals that the 2022 conflict in Northeast Ethiopia inflicted both direct and indirect health harms ranging from casualties, displacement, and violence to systemic collapse of healthcare services and long-term issues like PTSD and disability. Economically, the conflict has led to a significant reduction in tax revenues and an increase in military expenditures, straining the nation’s fiscal resources and hindering development efforts [ 35 ] . Recent studies by Muhyie et al. 2025 shows that armed conflict in the Amhara region devastated food systems, livelihoods, and social cohesion, displacing over 5.5 million people, destroying agricultural productivity, and causing infrastructure losses estimated at US $500 million. Additionally, Biset et al. 2023 finds that conflict in the region caused severe displacement (35%), high disease burden