Predicting Displacement: A Theoretical Critique of Data-Driven Forecasting of Forced Migration, Humanitarian Anticipation and the Politics of Early Warning

Abstract

Humanitarian agencies, governments and research consortia are investing in data-driven systems that forecast displacement from conflict, climate and economic signals. Their stated purposes are to position aid in advance and to plan reception. A less openly stated purpose is to prepare borders. The promise of anticipation is widely endorsed, yet the conditions under which displacement forecasting helps the people it predicts have received little theoretical attention. This conceptual, non-empirical paper offers a critique of displacement forecasting as both a scientific and a political practice. It integrates peer-reviewed scholarship published between 2023 and 2026 on migration and displacement forecasting, early warning and anticipatory action, the climate–conflict–displacement nexus, humanitarian data ethics and border externalisation, and every source was checked against its Crossref record and screened for retractions and editorial notices. From this synthesis the paper builds a predictable–actionable–beneficial framework, which distinguishes three claims that forecasting systems make: that displacement can be predicted, that prediction can be acted on, and that the action benefits the displaced. The analysis shows that each claim holds only under specific conditions. Forecasts tend to be most accurate where data are dense and contexts stable, which is often where they are least needed. A forecast that tells a humanitarian agency where to pre-position food can equally tell a border agency where to deploy patrols. Forecasting also renders displaced people as flows to be anticipated, although their own decisions to stay, flee or comply with armed actors shape the movements being predicted, and may be altered by the forecast itself. Five propositions follow, on accuracy under data sparsity and conflict, the institutional conditions of actionability, dual use, reflexivity and the governance of beneficence. The Lake Chad Basin and the central Sahel serve as the anchoring context. The paper concludes with an ethical framework for forecasting people who have not consented to being predicted, built on purpose limitation, a prohibition on use for containment, accountability to affected populations and independent oversight.