Climate refugees


· 9 min read
Climate refugee sounds like a stamp. The 1951 Refugee Convention does not create one. A European Parliament Research Service briefing, updated in June 2026, records the same point: there is no agreed definition, and climate is not, on its own, a ground for refugee protection. The briefing is research for members, not an adopted European law. The EU has not created the status.
I started from that phrase and followed the money. Science already maps the hazards. Governments already pay for disaster recovery. Insurance already prices some of the damage. Emissions trading already raises public revenue for climate policy. When people can no longer live where they lived, who pays for the move, and who is financially responsible for rebuilding a life rather than writing down a ruined house? A flood or a fire is not, by itself, proof of climate attribution. A recorded displacement is not a unique person, and it is not automatically a permanent relocation.
A hazard is the physical event. A climate risk is what happens when that hazard meets people, buildings, livelihoods and the capacity to respond. Displacement can last a night, years, or become a finding that return is no longer possible. Those three sit in the same tables. They are not the same obligation.
The IPCC Sixth Assessment treats drought, tropical storms, heavy rain and flooding as drivers of involuntary movement, and separates a house destroyed by a cyclone from a livelihood that slowly fails. Most of that movement stays inside national borders. The UN Office for Disaster Risk Reduction keeps the vocabulary that stops an earthquake being counted as climate evidence.
In March 2024 the European Environment Agency published the first European Climate Risk Assessment: 36 major climate risks in five clusters. Eight were judged particularly urgent. One of those files is the viability of European solidarity mechanisms, including the EU Solidarity Fund.
In September 2024 Storm Boris put extreme rain onto Central and Eastern Europe. The Internal Displacement Monitoring Centre’s Global Report on Internal Displacement 2025 recorded more than 72,000 internal displacements across Austria, the Czech Republic, Italy, Moldova, Poland, Romania and Slovakia. Almost 44,000 were in Poland, nearly all evacuations in the city of Nysa. More than 21,000 were recorded in the Czech Republic. These are movements in a year, not a census of people who permanently left.
On 27 May 2025 the Commission proposed about €280 million from the EU Solidarity Fund for the autumn 2024 floods, including Storm Boris and later flooding in Bosnia and Herzegovina. On 9 July 2025 the European Parliament and the Council adopted Decision (EU) 2025/1525, recording about €280 million as budget appropriations, not as money already paid out. The largest allocations were for Czechia, Poland and Austria.
The Fund helps states with public emergency and recovery spending: infrastructure, temporary accommodation, protective works, clean-up. It is not designed to compensate private losses. Temporary accommodation, in the regulation, lasts until people can return to their original homes after repair. Relocation or improvement of infrastructure can be eligible, capped at the cost of restoring the previous condition. The Fund does not decide whether a repeatedly flooded household should go back, be protected in place, or be offered a way out.
Le Signal was a residential building of 78 apartments at Soulac-sur-Mer on the French Atlantic coast. Progressive dune erosion made it unsafe. Evacuated in 2014, demolished in 2023. Occupation ended.
French disaster insurance and the Barnier fund, which can support expropriation for certain major natural risks, did not give those owners a clear route. Dune erosion was treated as foreseeable. The Council of State confirmed in 2018 that the case did not fit the Barnier criteria. The building could be declared uninhabitable while the financial treatment of the owners remained a separate problem.
A statute of 30 July 2020 let eligible owners claim 70% of the market value calculated as if the collapse risk were not there, if they transferred the property to a public body and waived related claims. The law covers principal and secondary homes. Most of the apartments were holiday homes. The state could see the risk. Compensation still needed a one-off statute.
During September to November 2024 the Philippines experienced an exceptional sequence of tropical cyclones, including six that struck Luzon in rapid succession. The World Meteorological Organization reported more than 13 million people affected across 17 of 18 regions, and more than 1.4 million displaced, across that broader sequence. The first financial problem is warnings, evacuations and emergency support, then rebuilding in places that may be hit again.
The country already had a national disaster-risk system. Republic Act 12287, approved on 12 September 2025, creates a way to declare a State of Imminent Disaster on forecasts and pre-disaster assessment, so that money and evacuation can move before impact. It was not in force during those storms. The implementing rules were published on 29 September 2026. That is financing protection before the roof comes off. It is not a system that pays for permanent resettlement when the same coast is no longer tenable.
Canada’s 2023 wildfire season burned about 15 million hectares. The UN disaster-risk office reports around 232,000 people evacuated. Heat, drought and fire weather belong in that season, together with ignition and land management.
Canada has insurance and a federal disaster-financing system, the Disaster Financial Assistance Arrangements, which cost-shares eligible provincial and territorial spending. It does not pay households directly and often will not duplicate insurable fire damage. The modernised guidelines apply to disasters beginning 1 April 2025. They must not be read backwards onto 2023. The newer design says the programme is not meant to restore assets that were high-risk and untenable over the long term. In defined circumstances it can support eligible buyouts and relocation.
Most of the people in those four files never crossed a border. UNHCR treats the phrase climate refugee as inaccurate and misleading. Under the 1951 Convention, refugee status generally requires crossing an international border because of a well-founded fear of persecution on specified grounds. Climate-related harm alone does not establish that status, although it may intersect with persecution or conflict.
In 2024 IDMC recorded 45.8 million disaster displacements, the highest in its series. About 45.5 million were weather-related. That is movements, including repeat evacuations. The same person can be counted more than once. At the end of that year, 9.8 million people were still living in displacement because of disasters. The first number is traffic. The second is people who had not gone home. People who stay inside their own country are waiting for a finance rule, not for Geneva.
Property insurance pays, where the peril is covered, for a qualifying loss after it happens. It prices a building. National disaster funds and the Solidarity Fund repay public emergency costs. France’s natural-disaster insurance is not a vacuum. The Signal case is a specific exclusion: a slow, foreseeable retreat of dunes did not open the ordinary door.
The Philippines can, after 2025, move money on a forecast. Canada’s revised DFAA guidelines can support eligible buyouts and relocation. A funded property transaction is not a guarantee of restored income, employment, community ties or long-term housing security.
In 2025, auctions under the European emissions trading system raised about €43 billion. Member states must use those revenues, or an equivalent amount, for climate and energy purposes, which can include adaptation. The money is already assigned. It is not a household entitlement to relocate when return is no longer viable.
This investigation did not find, in these four programmes, a published long-term follow-up of household welfare, housing stability, employment and remaining exposure after people were moved. Where that file is missing, the limitation is the evidence.
The proposal is an architecture of financial responsibility, not a priced mechanism, a European Union programme or a United Nations fund. It would still need modelling of sufficiency, risk-sharing and contribution rates. One path. Three inflows. Four moments at which that path would have to pay.
Government: public reserves, guarantees, official risk assessment, and the duty for essential infrastructure. Residents: a broad compulsory pool linked to property and to official climate-risk maps, with adjustments so that poorer households are not priced out. A fund fed only by the most exposed streets would be unstable the first time a whole region is hit at once. Emissions-related revenue: a share, set in law, of climate money that already exists, including auction proceeds, rather than assuming a brand-new carbon tax. The four stages are when the path pays. The three inflows are how it is filled.
Stage one, a recognised climate risk: the map has spoken, the water has not. Government pays for assessment and public defence; the pool can pay for household-level protection; emissions revenue can thicken prevention. Stage two, an imminent threat: government and the pool pay to get people out. That is the logic the Philippines wrote into statute after 2025. Stage three, verified damage: insurance and public compensation price a wall, a roof, a public pipe. Most existing machines live here. They stay as the third trigger, not the first and not the last. Stage four, a finding that return is not viable: the object of payment changes. Another place to live, and the conditions of a life. Government has to make a legal decision, not only a grant. The pool has to be broad enough to buy relocation without bankrupting the street that is leaving. Under this proposed framework, earmarked emissions revenue would support long-term relocation and livelihood recovery beyond insured property losses.
Recognising that an area is high-risk is not the same legal act as ordering a particular property abandoned. Residents would need a right to take part in that decision and to challenge the assessment.
Insurance and emergency funds are good at a building: market value, cost of repair, temporary beds, a road put back. Compensating destroyed property can close a claim file. It does not restore income, work, or an address less exposed than the last one.
None of the four machines, on the public record assembled here, is a monitored restoration of living standards. Governments can already identify where climate risks are growing. Existing systems can finance prevention, evacuation and reconstruction. What remains less clearly established is a continuous financial obligation to restore a household’s living conditions when returning home is no longer viable.
illuminem Voices is a democratic space presenting the opinions of leading Sustainability Thought Leaders, their views do not necessarily represent those of illuminem.
The world needs sustainability knowledge. At illuminem, no interest group or shareholder can influence our work. Thank you for supporting our mission to make high-quality and independent sustainability information free for all. Every contribution helps. Thank you for donating today.
Alex Byelyavtsev

Climate Change · Public Governance
illuminem briefings

Climate Change · Environmental Sustainability
illuminem briefings

Climate Change · Public Governance
Deutsche Welle

Climate Change · Public Governance
Associated Press

Climate Change · Public Governance
France24

Climate Change · Environmental Sustainability