When the Killing Slows but Displacement Keeps Growing

Two of the most-watched welfare signals in Factrail's security domain moved in opposite directions across 2022 and 2023, and the gap between them is the story.

Global conflict deaths, tracked through Global Deaths in Armed Conflicts (UCDP), peaked at roughly 401,500 in 2022 and then fell to about 149,900 in 2023 — still far above the model's ~50,000/year normal-line baseline, but a sharp year-on-year decline. Over the same window, People Internally Displaced by Conflict and Violence (IDMC) did the opposite, rising from 62.5 million at end-2022 to a record 68.3 million at end-2023.

What happened, and what is verified. The 2023 fall in deaths is well documented. A large part of it reflects the AU-brokered Pretoria Agreement that ended the Tigray war, signed on 2 November 2022 under African Union mediation. UCDP best-estimate totals compiled by Our World in Data confirm the deaths decline, and the 2024 Global Report on Internal Displacement confirms the record displacement figure of 75.9 million total IDPs, of whom 68.3 million were displaced by conflict and violence. Both numbers are high-confidence and source-backed.

Why the signals diverge. In Factrail's graph, both indicators sit downstream of the same pressure — Armed-Conflict Intensity — but through different mechanisms and time lags. The model links intensity to deaths with a near-immediate effect (a ~30-day lag, strong coupling), so when fighting stops, the death count responds within the same reporting year. The link from intensity to displacement carries a longer lag (~180 days) and, crucially, displacement is a stock, not a flow: people already uprooted remain counted as displaced long after the guns fall silent. The counter-pressure of Peacebuilding and Mediation Efforts eases deaths relatively quickly (~90-day lag) but only slowly reduces the displacement stock (modeled at a ~365-day lag with weaker strength), because return is hard, slow and often unsafe.

How this reads against the normal line. Deaths fell but stayed well above the ~50,000 baseline, so the indicator improved without returning to a calm state. Displacement, anchored to a ~41 million baseline around 2018, moved further from normal — a roughly 49% rise over five years per IDMC. So even the "good news" of fewer deaths coincided with a deepening structural deterioration in the displacement signal.

Alternative explanations. The divergence is not solely a lag artifact. New crises kept the displacement stock climbing even as the Tigray theatre cooled: IDMC attributes the 2023 record partly to Sudan, Gaza, Ukraine and the DRC. The Ukraine displacement crisis — Europe's largest since WWII — added millions to the global stock, and that fact is marked needs_review because wartime attribution is contested and is stated here only as documented by UNHCR. A second caveat: several pre-2022 displacement data points are reconstructed from IDMC's stated five-year change and carry medium confidence, so the precise shape of the 2018-2021 rise is approximate even though the 2022 and 2023 end-points are firm.

What may come next. If mediation holds and no comparable new war erupts, Factrail's structure implies deaths should stay below the 2022 peak while the displacement stock plateaus or eases only gradually — the slow-return dynamic means the IDP figure is unlikely to fall fast. That projection is developed in the domain forecast and is explicitly a scenario, not a fact.

Sources supporting the conclusion. The death series rests on UCDP totals via Our World in Data and the World Bank's republished battle-death series; the displacement figures rest on IDMC's GRID 2024, independently confirmed by the Norwegian Refugee Council. The Pretoria Agreement is documented by the African Union Peace and Security Department. Together they support the core finding: ending fighting cuts deaths faster than it reverses displacement.