Protracted Disaster Displacement: A Global Review of the State of Knowledge

About this Study

Hazardous and extreme events, such as earthquakes, storms and floods trigger large-scale evacuations and displacement every year. Pre-emptive and spontaneous evacuations for sudden-onset hazards respond to the immediate threat to life and are often associated with short-term displacement. Evidence is growing, however, that significant numbers of people remain displaced for long periods after such evacuations, or resettle permanently. Environmental degradation and slow onset processes and events further contribute to long-term displacement as sea-level rise, extreme temperatures and desertification result in the loss of habitable land and livelihoods.

In 2025, IDMC recorded 29.9 million internal disaster displacements and estimated that 13.6 million people remained displaced at the end of the year. Protracted displacement is often associated with negative impacts on the physical and mental wellbeing of individuals and with socioeconomic decline of households and communities. Thus, protracted disaster displacement is increasingly understood as a long, uneven process shaped by repeated hazard exposure, pre-existing vulnerability, housing, gender, and governance, rather than a temporary interruption that ends when the emergency response is over.

This study contributes to strengthening the evidence base on protracted disaster displacement and to informing policy, legal and financing responses aimed at preventing such situations and supporting durable solutions for affected populations. It specifically unpacks the factors prolonging displacement and hindering durable solutions and provides an overview of recent research and data on protracted displacement and displacement risk modelling. As such, it can be considered a snapshot of the state of knowledge on protracted displacement in the context of disasters.

Key Findings

• The scale of protracted displacement is not reliably measured:
  In 2025, 29.9 million internal disaster displacements were recorded and an estimated 13.6 million people remained displaced at year-end. Yet reporting on displaced populations typically ends within a week of sudden-onset events and within four months of slow-onset ones, so the point at which recovery stalls into protraction is largely invisible in global statistics.
• Definitions of “protracted” vary widely:
  Thresholds across the literature range from more than 12 months to more than five years. This review defines displacement as protracted when it has lasted at least 12 months, no durable solution (return, local integration, or settlement elsewhere) has been achieved, and displacement-linked vulnerability persists.
• Duration is itself a self-reinforcing risk factor:
  Across geographies and hazard types, the longer people remain displaced, the lower their likelihood of reaching a durable solution and the worse their physical, mental, and economic outcomes become. Evidence from post-tsunami Indonesia (Aceh), Japan, and post-earthquake Italy shows accumulating disadvantage over months and years of displacement, including elevated post-traumatic stress disorder (PTSD) symptoms persisting five to twelve years after the event.
• Physical damage is necessary but insufficient to explain protracted displacement:
  The scale of damage strongly predicts initial displacement, but socio-economic status, tenure, social networks, institutional capacity, and policy environment determine how long that displacement lasts. Remoteness, poverty, and population density compound the effects of housing and infrastructure damage.
• Homeownership and secure tenure are the most consistent household-level protective factors:
  Renters, squatters, informal occupants, and the landless face systematically higher barriers to return and to durable solutions across virtually every context studied. Land tenure emerges as the single strongest structural predictor of long-term housing quality and displacement resolution.
• The relationship between wealth and displacement duration is non-linear:
  The very poorest households can face immobility traps that prevent human mobility, middle-income households have the widest set of choices, and wealthier households can both move and return. Apparent contradictions across country studies dissolve once this non-linearity is accounted for.
• Slow-onset hazards can produce more persistent displacement than rapid-onset ones, however the evidence is not conclusive: Drought, heat stress, and environmental degradation drive long-term out-migration and permanent relocation in ways that established relief systems, which are largely oriented to sudden-onset events, do not buffer. Sea-level rise, coastal erosion, and desertification will increasingly render specific locations physically or economically uninhabitable, but adaptation policies can mitigate the risk significantly.
• Multi-hazard and conflict-affected contexts are qualitatively distinct:
  Where disasters intersect with conflict, protective factors such as tenure security and social capital operate very differently and can even be reversed. These contexts warrant separate analytical and policy treatment rather than extrapolation from single-hazard studies.
• Off-site, contractor-driven resettlement consistently produces poor long-term outcomes, and community-led, in-situ or near-site approaches show the strongest evidence of positive outcomes:
  Across Ecuador, the Philippines, China, Zimbabwe, Nepal, Sri Lanka, Bangladesh, and Iran, government-built off-site resettlement is associated with abandonment, declining satisfaction, disrupted livelihoods, and residents modifying or leaving units. Owner-driven and community-led reconstruction produce comparatively better outcomes but still under-serve the most marginalised.
• The separation of housing from livelihood support is the dominant mechanism driving poor outcomes:
  Programmes deliver shelter but not the economic infrastructure needed to sustain households in it. Access to land, employment, markets, and services near the new location repeatedly proves as decisive as the quality of the housing itself.
• Temporary housing routinely becomes long-term: Transitional shelter arrangements have persisted for years in Japan, Italy, Indonesia, and elsewhere. Prolonged residence in temporary housing directly drives population decline, mental health deterioration, and permanent out-migration.
• Governance and institutional design matter more than hazard magnitude in shaping displacement duration:
  Infrastructure recovery speed, coordination across agencies, legal recognition of internally displaced persons, perceived fairness of aid distribution, and community participation in decision-making are stronger determinants of outcomes than the physical intensity of the triggering event.
• Post-disaster programmes systematically produce differential outcomes:
  Renters, women, ethnic and indigenous minorities, extended and multi-generational families, people with disabilities, and older adults are consistently excluded from, or poorly served by, programmes nominally designed to help them.
• Uninhabitability is a socio-political condition, not a purely physical threshold: Projected ranges for climate-driven coastal migration this century vary by more than an order of magnitude, driven mainly by assumptions about adaptation policy, not climate uncertainty. Treating uninhabitability as inevitable risks undermining in-situ adaptation and legitimising forced displacement.
• Current risk models cannot yet estimate the duration of displacement:
  Probabilistic risk models now provide credible estimates of initial displacements associated with sudden-onset hazards, but they do not capture the socio-economic and institutional factors that determine whether displacement resolves in weeks or persists for years. No model currently answers the core question: given a hazard event of a given intensity, what share of the displaced remain so at 12 months, three years, or indefinitely?
• The evidence base is geographically and methodologically skewed:
  More than half of reviewed studies come from Sri Lanka, the Philippines, Indonesia, and Nepal, with limited coverage of sub-Saharan Africa, Latin America, and high-income island states. Most studies are qualitative and short-term, and treat displacement as a static state rather than an evolving process, limiting what can be said about long-term outcomes and durable solutions.
• A durable-solutions agenda requires moving beyond restoring pre-displacement conditions:
  Growing evidence supports investment that lifts displaced households to a higher level of resilience, akin to “building back better,” and explicitly integrates displaced populations into national development planning, social protection systems, and sectoral budgets, rather than treating displacement as a bounded humanitarian episode.

Report Contents

1. Introduction: Purpose and scope of the study, methodology
2. Key Findings: State of evidence, synthesis, and key gaps
3. The Limits of Knowledge: Understanding the scale of protracted disaster displacement
4. Displacement Pathways: Hazard and event characteristics, individual factors, household level factors, and contextual factors
5. Two Steps Forward, One Step Back: Solutions to protracted disaster displacement; housing programmes; governance, institutional design, and coordination; livelihoods and economic recovery; reflecting on solutions
6. Digging Deeper: Country case studies on Colombia (floods and institutional challenges), the Philippines (recurrent hazard exposure and compounded risk), and Somalia (land tenure insecurity and ongoing displacement)
7. A Glimpse of the Future: Assessing the risk of protracted displacement and understanding uninhabitability; what determines uninhabitability; who defines habitability?
8. Beyond the Usual Suspects: Recommendations for future research
9. References
10. Annexes: Annex 1 (Study Methodology) and Annex 2 (The Data Landscape on Protracted Disaster Displacement)

Author: Bina Desai | Published by the Platform on Disaster Displacement (PDD) and the Internal Displacement Monitoring Centre (IDMC)