UGA-001

AI-Enabled Refugee Forecasting for Uganda's Displacement Crisis Response Mechanism (DCRM)

Download PDF
Uganda Sub-Saharan Africa Low income Operational Deployment (Limited Rollout) Confirmed

Government of Uganda – Office of the Prime Minister (DRDIP); forecasting model developed by the World Bank.

At a Glance

What it does Prediction (including forecasting) — Trend and shock forecasting
Who runs it Government of Uganda – Office of the Prime Minister (DRDIP); forecasting model developed by the World Bank.
Programme Displacement Crisis Response Mechanism (DCRM) under DRDIP
Confidence Confirmed
Deployment Status Operational Deployment (Limited Rollout)
Key Risks Data-related risks
Key Outcomes First DCRM activation (2021 pilot): USD 1.
Source Quality 6 sources — Report (multilateral / development partner), News article / media, Working paper / technical note

Uganda's Displacement Crisis Response Mechanism (DCRM) is the world's first displacement risk financing mechanism, developed as a component of the Development Response to Displacement Impact Project (DRDIP) implemented by the Government of Uganda's Office of the Prime Minister. The DCRM integrates an artificial intelligence forecasting model developed by the World Bank to predict refugee inflows into Uganda from the Democratic Republic of Congo and South Sudan, enabling anticipatory public service scale-up in refugee-hosting districts before large population movements materialise.

The AI forecasting model is a machine-learning system that ingests over 90 independent variables spanning hundreds of dimensions across economic, social, natural, and built environments. Specifically, the model analyses conflict indicators, economic activity data, climate and vegetation monitoring data, food and market price information, built infrastructure assessments, and online language sentiment and volume regarding conflict, gender, and governance discourse drawn from news and social media sources. A distinctive feature of the model's design is its incorporation of perception-based information alongside concrete measurable data, reflecting the recognition that human displacement behaviour is driven not only by objective conditions but also by perceptions of change. The model was trained and tested using UNHCR daily refugee arrival data covering the period from 2014 to 2023, with separate models calibrated for arrivals from South Sudan and the Democratic Republic of Congo. When tested on unseen data, the model demonstrated over 80 percent accuracy in forecasting changes in future volumes of refugee inflows, producing predictions approximately four to six months ahead of actual arrivals.

The DCRM operates through a rules-based, pre-financed mechanism that prearranges contingency funding and pre-agrees intervention protocols before displacement shocks occur. A formal DCRM handbook, adopted by the Government of Uganda, outlines the guiding principles and operational processes for activation, including the use of need indices that measure the ratio of persons to public service facilities across refugee-hosting districts, identifying areas with the greatest per-capita service gaps. When the AI model forecasts significant refugee inflows or when actual inflows exceed pre-agreed thresholds, the mechanism is triggered and funds are disbursed to eligible host districts for rapid scale-up of public services. The activation process follows a transparent, needs-based, and pre-agreed procedure for sector and district selection, ensuring proportionate allocation based on documented needs rather than ad-hoc responses. Government officials retain decision authority throughout the disbursement process; the AI forecasts inform but do not determine funding decisions.

The DCRM has been activated twice. The first pilot activation occurred in 2021, disbursing approximately USD 1.2 million to two host districts. The second activation in 2023 was substantially larger, disbursing USD 3.3 million across all 13 eligible refugee-hosting districts, nearly triple the pilot amount. The 2023 activation prioritised water infrastructure investments in response to drought conditions that had placed additional stress on existing water sources in host communities. Both activations funded scale-up of public services in water access, health centre capacity, and classroom expansion. The mechanism also incorporates object classification models applied to satellite imagery for robust displacement impact assessment as part of its advanced data collection approach.

The DCRM is funded through the DRDIP project, which represents approximately USD 50 million in IDA financing from the World Bank to the Government of Uganda, with a pre-financed contingency allocation of USD 4.5 million dedicated to the DCRM component. Additional support has been provided by the World Bank-administered Multi-Donor Trust Fund for Forced Displacement and the PROSPECTS Partnership, with funding from the Kingdom of the Netherlands. The mechanism leverages Uganda's progressive refugee policies, which grant refugees rights to employment, land allocation, and access to national public services, integrating displacement response within the national development framework rather than treating it as a separate humanitarian exercise.

The initiative represents a shift from reactive to anticipatory humanitarian and development response. By enabling governments to build water points, expand health facilities, and construct classrooms before refugees arrive, the DCRM has contributed to reducing tension between refugee and host communities, strengthening community resilience, and enabling more efficient resource allocation. The World Bank has noted that the AI model is adaptable for predicting and explaining other development challenges beyond displacement, such as poverty levels and macro-fiscal pressures, suggesting potential for broader application of the forecasting approach within social protection systems.

Classifications follow the DCI AI Hub Taxonomy. Hover over field labels for definitions.

Social Protection Functions

Implementation/delivery chain
Assessment of needs/conditions + enrolment primaryMonitoring and evaluation Provision of payments/services
SP Pillar (Primary) The social protection branch: social assistance, social insurance, or labour market programmes. Social assistance
Programme Name Displacement Crisis Response Mechanism (DCRM) under DRDIP
Programme Type The type of social protection programme, classified under social assistance, social insurance, or labour market programmes. View in glossary Emergency Cash Transfers
System Level Where in the social protection system the AI is applied: policy level, programme design, or implementation/delivery chain. View in glossary Implementation/delivery chain
Programme Description The DCRM is a rules-based, pre-financed contingency funding mechanism embedded within the Development Response to Displacement Impact Project (DRDIP). It prearranges funds and pre-agrees intervention protocols to enable rapid public service scale-up in refugee-hosting districts when displacement thresholds are breached or forecast.
Implementation Type How the AI output is produced: Classical ML, Deep learning, Foundation model, or Hybrid. Affects validation, compute requirements, and governance profile. View in glossary Classical ML
Lifecycle Stage Current stage in the AI lifecycle, from problem identification through to monitoring, maintenance and decommissioning. View in glossary Monitoring, Maintenance and Decommissioning
Model Provenance Origin of the AI model: developed in-house, adapted from open-source, commercial/proprietary, or accessed via third-party API. View in glossary Developed in-house
Compute Environment Where the AI system runs: on-premise, government cloud, commercial cloud, or edge/device. View in glossary Not documented
Sovereignty Quadrant Classification of data and compute sovereignty: I (Sovereign), II (Federated/Hybrid), III (Cloud with safeguards), or IV (Shared Innovation Zone). View in glossary Not assessed
Data Residency Where the data used by the AI system is stored: domestic, regional, or international. View in glossary Not documented
Cross-Border Transfer Whether data crosses national borders, and if so, whether documented safeguards are in place. View in glossary Not documented
Decision Criticality The rights impact of the decision the AI supports. High criticality requires HITL oversight; moderate requires HOTL; low may operate HOOTL. View in glossary Moderate
Human Oversight Type Level of human involvement: Human-in-the-Loop (active review), Human-on-the-Loop (monitoring), or Human-out-of-the-Loop (periodic audit). View in glossary HOTL
Development Process Whether the AI system was developed fully in-house, through a mix of in-house and third-party, or fully by an external provider. View in glossary Mix of in-house and third-party
Highest Risk Category The most significant structural risk source identified: data, model, operational, governance, or market/sovereignty risks. View in glossary Data-related risks
Risk Assessment Status Whether a formal risk assessment, informal assessment, or independent audit has been conducted for this system. Not assessed

Risk Dimensions

Market, sovereignty and industry structure risks
Operational and system integration risks

Impact Dimensions

Accountability, transparency and redress
Systemic and societal
  • Exit/rollback plan
  • Human oversight protocol
CategorySensitivityCross-System LinkageAvailabilityKey Constraints
Administrative data from other sectorsNon-personalLinks data across multiple systemsCurrently available and usedUNHCR daily refugee arrival data (2014-2023); conflict, economic, climate, and food price data from multiple international sources; availability and timeliness may vary by source country (DRC, South Sudan)
Geospatial and remote sensing dataNon-personalLinks data across multiple systemsCurrently available and usedSatellite imagery for object classification models; climate and vegetation monitoring data; built infrastructure assessment data
Unstructured and text-based contentNon-personalSingle source (no linkage)Currently available and usedOnline language sentiment and volume from news and social media regarding conflict, gender, and governance discourse; subject to noise and language-specific limitations

Financial Protection Forum (2023). 'According to Plan: Second Activation of Uganda's Displacement Crisis Response Mechanism (DCRM)'. Washington, DC: World Bank SPJ Platform. Available at: https://www.financialprotectionforum.org/blog/according-to-plan-second-activation-of-ugandas-displacement-crisis-response-mechanism-dcrm (Accessed 24 March 2026).

View source Report (multilateral / development partner)

World Bank (2019). 'Machine Learning in Uganda Brings the Power of Risk Financing to Strengthen Refugee and Host Community Resilience'. Nasikiliza blog. Available at: https://blogs.worldbank.org/nasikiliza/machine-learning-uganda-brings-power-risk-financing-strengthen-refugee-and-host (Accessed 24 March 2026).

View source News article / media

World Bank (2020). 'Data-Driven Development Response to Displacement Crisis in Uganda: The Displacement Crisis Response Mechanism'. Washington, DC: World Bank. Available at: https://documents1.worldbank.org/curated/en/472101606119818621/pdf/Data-Driven-Development-Response-to-Displacement-Crisis-in-Uganda-The-Displacement-Crisis-Response-Mechanism.pdf (Accessed 24 March 2026).

View source Report (multilateral / development partner)

World Bank (2021). 'How AI Can Support Anticipatory Action to Address Forced Displacement'. Development for Peace blog. Available at: https://blogs.worldbank.org/en/dev4peace/how-ai-can-support-anticipatory-action-to-address-forced-displac (Accessed 24 March 2026).

View source News article / media

World Bank (2025). 'AI-Powered Refugee Forecasting: Preparing for Refugee Movements Before They Happen'. Washington, DC: World Bank. Available at: https://www.worldbank.org/en/topic/fragilityconflictviolence/brief/ai-powered-refugee-forecasting-preparing-for-refugee-movements-before-they-happen (Accessed 24 March 2026).

View source Report (multilateral / development partner)

World Bank (2025). 'AI-Powered Refugee Forecasting' [one-pager]. Washington, DC: World Bank. Available at: https://thedocs.worldbank.org/en/doc/4d816aaea9713abd30f63c6b92e1e79b-0090082025/original/AI-Powered-refugee-forecasting.pdf (Accessed 24 March 2026).

View source Working paper / technical note
Deployment Status How far the system has progressed into real-world operational use, from concept/exploration through to scaled and institutionalised. View in glossary Operational Deployment (Limited Rollout)
Year Initiated The year the AI system was first initiated or development began. 2018
Scale / Coverage The scale and geographic or population coverage of the deployment. 13 refugee-hosting districts in Uganda (2023 activation); model covers arrivals from South Sudan and DRC
Funding Source The source(s) of funding for the AI system development and deployment. IDA financing (DRDIP ~USD 50 million); DCRM contingency allocation USD 4.5 million; World Bank Multi-Donor Trust Fund for Forced Displacement; PROSPECTS Partnership (Kingdom of the Netherlands)
Technical Partners External technology vendors, academic partners, or development partners involved. World Bank (AI forecasting model development); UNHCR (refugee arrival data); PROSPECTS Partnership; Kingdom of the Netherlands (funding support)
Outcomes / Results First DCRM activation (2021 pilot): USD 1.2 million disbursed to 2 host districts. Second activation (2023): USD 3.3 million disbursed across all 13 eligible host districts, prioritising water infrastructure due to drought. AI model demonstrated over 80% accuracy on unseen data for forecasting refugee inflow volumes 4-6 months ahead. Outcomes include reduced tension between refugee and host communities, strengthened community resilience, and more efficient anticipatory resource allocation for water, health, and education services.

How to Cite

DCI AI Hub (2026). 'AI-Enabled Refugee Forecasting for Uganda's Displacement Crisis Response Mechanism (DCRM)', AI Hub AI Tracker, case UGA-001. Digital Convergence Initiative. Available at: https://socialprotectionai.org/use-case/UGA-001 [Accessed: 1 April 2026].

Change History

Updated 1 Apr 2026, 08:11
by system (system)
Created 30 Mar 2026, 08:41
by v2-import (import)