Patterns in conflict dynamics in Yemen and Syria

Conflict fatalities tend to follow heavy-tailed statistical distributions. A 2005 fusion-fission theory predicts that, when armed groups operate through dynamically forming and fragmenting clusters, the number of fatalities per event should be approximated by a power-law distribution with exponent $\alpha$ near 2.5, with deviations reflecting the relative robustness of larger versus smaller fighter clusters. We use Armed Conflict Location and Event Data (ACLED) event-level fatality data to estimate $\alpha$ in recent Yemen and Syria conflict periods during 2023-2025. These conflict event datasets are small, incomplete and imperfect and only produce very crude approximations to a power law; nonetheless, the best-fit $\alpha$ values occupy the range $2.5<\alpha <3.5$ across most periods, which is broadly consistent with predominantly fragmented or weakly consolidated armed-group organization. Temporary reductions in $\alpha$ are broadly consistent with episodes in which larger coordinated clusters become more consequential. More rigorous testing will require improved datasets to become available in the future.