In this research, we aim to identify the main factors that explain the occurrence and intensity of armed conflicts in a specific region, the Middle East and North Africa. We extend the conventional linear Bayesian Model Averaging procedure by incorporating conflict intensity, which is measured across a spectrum of violence levels, departing from the typical binary classification of war or peace. We provide strong evidence that not only demographical, institutional and socio-economic but also, environmental factors must be considered when analyzing conflict intensity. By paying special attention to neighboring states’ characteristics, our results reveal that political economy factors, historical legacy, climate and access to natural resources are key in identifying conflict severity. Finally, we show that model averaging predictions for ordered categorical outcomes improve upon the existing out-of-sample conflict prediction techniques.
Authors
Olivier Parent
Professor, Department of Economics, University of Cincinnati
Research Fellows
Abdallah Zouache
Full Professor in Economics, Sciences Po Lille,...