# The Role of Governmental Weapons Procurements in Forecasting Monthly Fatalities in Intrastate Conflicts: A Semiparametric Hierarchical Hurdle Model

**Authors:** Cornelius Fritz, Marius Mehrl, Paul W. Thurner, Göran Kauermann  
**arXiv:** [2012.08246](https://arxiv.org/abs/2012.08246)  
**Date:** 2020-12-15  
**Summary page:** https://www.summarizepaper.com/en/paper/2012.08246v1/

## Key points

- Study title: "The Role of Governmental Weapons Procurements in Forecasting Monthly Fatalities in Intrastate Conflicts: A Semiparametric Hierarchical Hurdle Model"
- Authors: Cornelius Fritz, Marius Mehrl, Paul W. Thurner, and Göran Kauermann
- Need for accurate forecasting models to predict changes in intrastate conflict casualties
- Proposal of a hierarchical hurdle regression model operating at the monthly PRIO-grid level
- Conceptualization of local armed conflicts as a three-stage process
- Examination of governmental arms imports as a key factor influencing conflict escalation or de-escalation
- Consideration of geographic remoteness of grid cells in moderating effects of military buildups on conflict outcomes
- Effectiveness of the model demonstrated through out-of-sample predictions
- Emphasis on transparency and high accuracy in predicting complex events
- Contribution to understanding and forecasting intrastate conflicts with practical implications for decision-makers and humanitarian organizations

## Summary

In their study titled "The Role of Governmental Weapons Procurements in Forecasting Monthly Fatalities in Intrastate Conflicts: A Semiparametric Hierarchical Hurdle Model," authors Cornelius Fritz, Marius Mehrl, Paul W. Thurner, and Göran Kauermann address the critical need for accurate and interpretable forecasting models to predict spatially and temporally fine-grained changes in intrastate conflict casualties. These forecasts are essential for informing policymakers and international non-governmental organizations (NGOs) on conflict dynamics. The researchers propose a hierarchical hurdle regression model that operates at the monthly PRIO-grid level to tackle the prediction challenge effectively. This model conceptualizes the intensity of local armed conflicts as a three-stage process. The first two stages estimate the likelihood of observing casualties at the country- and grid-cell-levels, respectively. In contrast, the third stage employs a regression model for truncated data to predict the number of fatalities based on the outcomes of the previous stages. Central to their modeling framework is an examination of governmental arms imports as a key factor influencing conflict escalation or de-escalation. The authors argue that these weapons procurements play a significant role in shaping conflict dynamics. Additionally, they suggest that geographic remoteness of grid cells can moderate the effects of military buildups, highlighting how spatial factors influence conflict outcomes. Through out-of-sample predictions, Fritz et al. demonstrate the effectiveness of their parsimonious and theory-driven model in accurately forecasting intrastate conflict casualties. Their approach emphasizes transparency while maintaining high levels of accuracy in predicting these complex events. Overall, this research contributes valuable insights into understanding and forecasting intrastate conflicts, offering practical implications for decision-makers and humanitarian organizations working in conflict-affected regions.