Bayesian Projection of Refugee and Asylum Seeker Populations

2024-05-11

AI Summary

This study addresses the challenge of long-term population forecasting for refugees and asylum seekers. We propose a Bayesian hierarchical time-series model grounded in crisis lifecycle theory, applied to UNHCR official data covering 35 protracted crisis countries. For the first time, we embed an interrupted logistic growth–decay process into forced migration forecasting, explicitly capturing the three-phase dynamic evolution: growth, peak, and decline. The Bayesian hierarchical structure enables cross-country information sharing and rigorous uncertainty quantification. Our model achieves robust and superior performance across multi-horizon forecasting windows—1-year, 5-year, and 10-year—delivering interpretable, updateable, and uncertainty-annotated medium- to long-term population projections. This provides policymakers with a principled, evidence-based tool for strategic planning and resource allocation in humanitarian response.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Life-Long and Continual Learning
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Models for Web evolution
• User Modeling, Personalization and Recommendation: Studies of user behavior, including longitudinal effects of personalized systems

Abstract

Estimates of future migration patterns are a crucial input to world population projections. Forced migration, including refugee and asylum seekers, plays an important role in overall migration patterns, but is notoriously difficult to forecast. We propose a modeling pipeline based on Bayesian hierarchical time-series modeling for projecting combined refugee and asylum seeker populations by country of origin using data from the United Nations High Commissioner for Human Rights (UNHCR). Our approach is based on a conceptual model of refugee crises following growth and decline phases, separated by a peak. The growth and decline phases are modeled by logistic growth and decline through an interrupted logistic process model. We evaluate our method through a set of validation exercises that show it has good performance for forecasts at 1, 5, and 10 year horizons, and we present projections for 35 countries with ongoing refugee crises.

Problem

• Projecting refugee population statistics using Bayesian modeling
• Forecasting forced migration patterns for multiple time horizons
• Modeling refugee population growth and decline phases

Innovation

• Bayesian hierarchical time-series modeling for refugee projections
• Interrupted logistic process modeling growth and decline phases
• Validated forecasting at one five and ten year horizons