After Repatriation Died

Executive Summary. Bangladesh's registered Rohingya population crossed one million in 2024 (UNHCR: 1,005,602 as of 31 December 2024), a milestone Bangladesh's policy institutions have not formally marked. The repatriation framework that governed policy since 2017 has no operative content: two pilot operations in 2018 and 2019 moved zero people, Myanmar's Arakan Army now controls most of Rakhine State, and the civilian junta cannot enforce return commitments. WFP cut the monthly food ration to $6 per person in April 2025, the lowest value since the response began, half the twelve-dollar standard ration and below the eight-dollar "subsistence floor" WFP itself named in 2023. As of May 2026, Foreign Minister Dr. Khalilur Rahman has stated the BNP government will pursue repatriation drawing on past successes in 1978 and 1992, but acknowledged that active conflict in Myanmar prevents any near-term movement. The brief argues for five administrative measures, none requiring new legislation, to shift the framework from indefinite containment toward managed coexistence.

In August 2017 and the months that followed, more than seven hundred and forty thousand Rohingya crossed the Naf River from Rakhine State into Bangladesh. The arithmetic of that crossing has dominated the public conversation for nine years: how many came, how fast, what Myanmar's military did, what the international community said, and when they would go back. By the end of 2024, the World Bank's reproduction of the UNHCR Population Statistics Database recorded one million five thousand six hundred and two persons under UNHCR mandate in Bangladesh, by country of asylum. The UNHCR-Government of Bangladesh joint factsheet for 31 December 2024 gave a near-identical figure of 1,005,520 across 204,278 families. The number is not a stock waiting to drain. It is a population that grew through 2024 from natural increase, that grew further from the 2024 to 2025 surge of new arrivals fleeing Arakan Army advances in northern Rakhine, and that has now produced more children inside the camps than the original adult arrivals brought with them.

The number has crossed a threshold and almost nobody has marked it. The 1991 census of Bangladesh recorded only two divisions of the country, Sylhet and Barishal in their later forms, with total populations under ten million each. The Rohingya population in Cox's Bazar today is larger than the entire population of Khagrachhari and Bandarban districts combined, larger than the city corporation of Rajshahi, larger than the Bangladeshi populations of every district in the Hill Tracts together. The policy framework Bangladesh has used to govern this population since 2017 has one assumption at its core: it is temporary. The data say otherwise.

This brief argues that the public conversation about the Rohingya in Bangladesh is anchored to a counterfactual that has not been operationally true since 2018, and that the resulting policy vacuum is now the most consequential cost. The counterfactual is repatriation. The reality is permanence. The cost of pretending otherwise is borne by the Rohingya, by the host community in Ukhia and Teknaf, and by Bangladesh's own fiscal trajectory through 2030. We use the verb "borne" rather than "paid" deliberately. Most of these costs do not appear in any line of the national budget. That is part of the problem.

What the data actually show

The starting point is the World Bank reproduction of UNHCR's Population Statistics Database. The series SM.POP.RHCR.EA records the stock of refugees under UNHCR mandate in a country of asylum at year-end. For Bangladesh:

Year	Refugees in BD	Year-on-year
2014	232,468	+ 1,326
2015	231,954	- 514
2016	276,203	+ 44,249
2017	932,209	+ 656,006
2018	906,640	- 25,569
2019	854,779	- 51,861
2020	866,534	+ 11,755
2021	918,907	+ 52,373
2022	952,384	+ 33,477
2023	971,984	+ 19,600
2024	1,005,602	+ 33,618

Three points need to be made about this table.

First, the 2017 figure is the registered count, not the situation report count. UNHCR situation reports from late 2017 referred to seven hundred and forty thousand new arrivals in eight weeks. The registration backlog was substantial through 2018, which is why the year-end registered total fell from 932,209 to 906,640 between 2017 and 2018: people were being deduplicated, not departing.

Second, since 2018 the registered population has not declined in any single year. The brief drops in 2018 and 2019 reflect registration cleanup. The flat to slightly rising trajectory from 2020 onward reflects natural increase. With a population of roughly one million and a crude birth rate consistent with displaced populations from neighbouring South Asia, the camps would generate around thirty thousand to forty thousand births per year against a much smaller mortality count. The 33,477 net increase from 2021 to 2022 is arithmetically consistent with natural increase alone. The 33,618 net increase from 2023 to 2024, and the further increase that the 2025 figures will show when finalised, also include a wave of new arrivals from Maungdaw and Buthidaung as the Arakan Army advanced into northern Rakhine. Bangladesh registered approximately 120,000 new arrivals in the camps from May 2024 onward, a headcount approved in October 2024 found 65,000 individuals, and total estimates for 2024 to 2025 cumulative new arrivals are in the range of 150,000 to 200,000. The 2024 to 2025 inflow is the largest since 2017 and has been almost entirely absent from the public conversation in Bangladesh.

Third, the registered population crossed one million in 2024. Bangladeshi government officials, ministry of foreign affairs spokespeople, and most public commentary continue to use the round figure of "1.2 million Rohingya," which adds an estimate of unregistered arrivals from earlier exodus waves to the registered count. The estimate is plausible and is consistent with what Bangladesh has communicated to international donors. But the registered, UNHCR-mandated number, which is what donor commitments and joint response plans are sized to, has just crossed seven figures. That is itself a milestone worth marking, and Bangladeshi institutions have not marked it.

The counterfactual that is not coming back

Repatriation has been the assumed exit ramp since the first arrivals. The 1992 caseload, in fact, mostly went home, by 1997 over two hundred thousand of the 1991-1992 arrivals had returned to Rakhine under a UNHCR-monitored arrangement and one of the legacy refugee camps at Kutupalong dwindled to fewer than thirty thousand by 2014. That precedent shaped policy thinking in the immediate aftermath of August 2017. The November 2017 bilateral arrangement between Dhaka and Naypyitaw envisaged returns beginning in early 2018. Two pilot operations were attempted, in November 2018 and August 2019. Neither moved a single person. The Rohingya themselves refused to return, citing the absence of citizenship guarantees, the destruction of villages, and the continuing presence of the same military commanders who had directed the 2017 operation.

Three things have happened since that have made the original counterfactual obsolete.

The Myanmar civil war that began after the February 2021 coup has destroyed the operational possibility of organized return. The Arakan Army, which was already a significant force in northern Rakhine in 2017, took effective control of the state-internal security apparatus in 2024 and now governs most of the territory the Rohingya were expelled from. The State Administration Council in Naypyitaw has lost the ability to make commitments about Rakhine that it can enforce. The Arakan Army has made statements about Rohingya return that range from neutral to actively hostile, and has been credibly accused by Human Rights Watch and Fortify Rights of separate atrocities against Rohingya remaining in Buthidaung in 2024. Whatever Bangladesh negotiates with whichever entity controls Rakhine, the receiving conditions have deteriorated, not improved.

The international diplomatic apparatus that pressed Myanmar between 2017 and 2021 has dispersed. The Gambia's case at the International Court of Justice, accepted as plausible in 2020, is moving but will not produce return as a remedy. The UN Independent Investigative Mechanism for Myanmar continues to compile evidence for future prosecutions but lacks enforcement authority. The ASEAN Five Point Consensus has not been implemented. The Special Envoy of the UN Secretary-General for Myanmar has had three holders in five years, none of whom have produced a workable framework for return.

The Bangladesh political transition of 2024 disrupted the bilateral channel that had sustained intermittent contact with Naypyitaw. The interim government has made statements supporting return but has not initiated new bilateral negotiations. The next elected government will face the same structural problem: there is no entity in Myanmar with the territorial control and the political legitimacy to receive a million people back into a region where the receiving infrastructure was destroyed and the receiving population is itself displaced by ongoing fighting.

This is not a temporary delay. It is a structural change in the policy environment. The repatriation framework that organized Bangladeshi thinking from 2017 to 2023 has run out of operative content. What remains is a policy framework whose stated purpose, organized return, has no realistic timeline, and whose actual effect, indefinite containment, has no honest accounting.

The funding curve nobody is fixing

The financial arrangement Bangladesh accepted in 2017 was that the international community would underwrite the operational costs of hosting the Rohingya through annual Joint Response Plans coordinated by UNHCR and IOM. The plans set a yearly appeal, donors pledge, and the funded share covers food, shelter, water, sanitation, education, health care, protection, and the operating costs of the camp administration. Bangladesh provides land, security, and registration services and absorbs the indirect costs to the host community.

The arrangement has held in form for nine years. In substance it has been deteriorating since 2022. The 2018 plan was funded at 67.9 percent. The 2019 plan reached 74.4 percent. The 2020 plan, distorted by the global pandemic, fell to 57.9 percent. The 2021 plan recovered to 73.4 percent. Then the curve broke. The 2022 plan was funded at 60.0 percent. The 2023 plan was funded at 50.7 percent, the lowest annual share since the response began. The 2024 plan partially recovered to 64.3 percent, with $548.2 million received against a $852.4 million appeal as of 23 January 2025. The 2024 figure is misleadingly stable, however, because the underlying need rose during 2024 with the arrival of approximately one hundred and twenty thousand new refugees from Rakhine, who were not in the original 2024 appeal at all. For 2025, the Joint Response Plan moved to a biennial framework: $934.5 million for the eighteen months covering January 2025 to December 2026, of which approximately $84 million was budgeted specifically for the up-to-150,000 new arrivals expected during the plan period. As of mid-2025 the new arrivals component was only partially funded and the response framework as a whole had a more than $176 million first-priority gap.

The funding deterioration is not a coincidence of donor reluctance about Bangladesh. It is the structural redistribution of humanitarian funding away from protracted situations toward acute ones. Ukraine in 2022 absorbed nearly thirty billion dollars of incremental humanitarian and reconstruction funding from the same donor pool that had previously funded the Bangladesh response at scale. Gaza and Sudan in 2024 added comparable demand. The Rohingya response is not failing on its own merits. It is being squeezed by a humanitarian financing system that was not designed to sustain a permanent population for a decade.

The most visible cost of the funding deterioration has been the food ration. WFP began the response in late 2017 at a value of approximately ten dollars per person per month, increased to twelve dollars during the standard ration period from 2019 through 2022, and held there through the COVID years. In March 2023 WFP cut the per-person value from twelve dollars to ten. In June 2023 WFP cut the per-person value from ten to eight, and used the phrase "subsistence floor" in the public statement, an unusually direct admission that the ration could not be cut further without crossing into food insecurity by WFP's own definitions. A partial restore to ten dollars came in December 2023, then a full restore to twelve and a half in January 2024 after a US emergency contribution, then a hold through August 2024.

In April 2025, WFP cut the per-person value from twelve dollars and fifty cents to six dollars, effective 1 April. The cut was announced in a WFP statement on 5 March 2025 as a contingency measure given the funding outlook for the 2025-26 plan. The reduction of six dollars and fifty cents per person per month, applied across roughly one million beneficiaries, removed approximately seventy-eight million dollars per year of food purchasing power from the camps in a single decision. As of the date of this brief in May 2026, the ration has been held at six dollars for thirteen months, the longest sustained ration period in the history of the response and the lowest dollar value since the operation began.

The 2023 round of cuts to ten dollars and then eight dollars produced documented nutrition consequences. The Cox's Bazar Nutrition Surveillance round 2 in September 2023 measured the global acute malnutrition prevalence in children under five at fifteen point one percent in Camp 24 and similar levels across the broader response, exceeding the World Health Organization fifteen percent emergency threshold for the first time since the original 2017 emergency phase. Maternal anaemia and exclusive breastfeeding indicators all moved in the wrong direction. The 2025 ration of six dollars falls below even the 2023 "subsistence floor" of eight dollars. The nutrition survey rounds for 2026 will measure the consequences. The direction is not in doubt.

Six dollars per person per month is the cost of approximately five kilograms of rice and a small allocation of pulses and oil at Cox's Bazar wholesale prices. It is below the WFP minimum food basket for the camp setting. It does not include any allocation for fresh vegetables, animal protein, micronutrient-rich food, or any non-food essential. The IPC Phase Classification system that WFP itself uses to categorize food insecurity would rate a population sustained on this ration over a year or more as Phase 3 (Crisis) or Phase 4 (Emergency) depending on the supplementary feeding available. UNHCR field reports from late 2025 documented increasing rates of acute malnutrition in children under five in the camps and a measurable increase in maternal anaemia.

This is the operational situation in which Bangladesh's policy framework has nothing to say. The framework assumes the funding will recover when international attention returns. The donor data say it will not.

The host community arithmetic

The Cox's Bazar district had a population of roughly two point three million in the 2011 census, projecting to approximately two point seven million by 2017 at the district's pre-influx growth rate. The August 2017 arrivals concentrated in two upazilas, Ukhia and Teknaf, whose combined pre-influx population was approximately five hundred thousand. The arrival of seven hundred and forty thousand Rohingya into these two upazilas, plus the longer-tenured caseload, produced an upazila-level demographic inversion the modern history of South Asia has no parallel for. Ukhia and Teknaf became, overnight, majority-Rohingya territories in which the original Bangladeshi population was a minority by a factor of roughly two to one.

The economic consequences for the host community are documented in the World Bank's 2019 and 2022 Cox's Bazar Panel Surveys, in BIDS' host community labour market studies from 2020 and 2023, and in IFPRI's 2021 wage and food price work. The findings, broadly converging across these studies, are:

Daily wages for unskilled male agricultural and non-agricultural labour in Ukhia and Teknaf fell relative to the rest of Cox's Bazar district by ten to fifteen percent in the eighteen months after August 2017, then partially converged but remained five to eight percent below the district trend through 2022. The wage gap was largest for the most casualized work, which is precisely the segment that competes most directly with refugee labour offered through informal arrangements that bypass the official prohibition on Rohingya formal employment.

Land and rental prices in Ukhia and Teknaf moved in the opposite direction. Rents for housing and small commercial space rose by twenty-five to forty percent above the district trend, driven by demand from international NGO staff, local NGO staff, and the supply chain that grew up around the camp economy. The aggregate effect was a redistribution within the host community: workers lost, landlords gained, traders gained, and the food-insecure poor lost most.

The forest cover loss in the camp area is one of the few host-community impacts where the data are unambiguous and the responsibility traceable. Between September 2017 and the end of 2018, satellite imagery analysis by UN-Habitat, FAO, and the German aerospace centre DLR documented the loss of approximately six thousand five hundred hectares of natural forest in the Ukhia-Teknaf belt, primarily for shelter construction and cooking fuel. The loss has not been replaced. The Liquid Petroleum Gas distribution programme funded by IOM from 2019 onward did substitute for fuelwood in approximately seventy percent of camp households by 2022, but the original forest loss is permanent and the regenerated areas are degraded scrub.

The water table in Ukhia is the host-community impact that will become more visible in the coming five years. The camp population draws roughly fifteen thousand cubic metres per day from shallow tubewells. Groundwater monitoring by DPHE and UNHCR's water and sanitation cluster has documented progressive lowering of the static water level in the camp belt, with measurable effects extending into adjacent host community areas. The longer-term solution, a piped surface water supply from the Bakkhali river basin, has been costed at approximately ninety million dollars and is half-funded.

The aggregate effect on Cox's Bazar as a district is harder to summarize because the camp economy has also generated income for hosts. The World Bank's 2022 panel found that the median host household in Ukhia and Teknaf had a higher real consumption level in 2022 than in 2017, but with greater dispersion: a substantial minority of host households were materially worse off. Aggregate poverty in Ukhia upazila, measured against the upper poverty line, was higher in 2022 than in 2016. The camp economy has not lifted the host community out of poverty. It has reorganized it.

The Bhasan Char miscalculation

The Bhasan Char relocation was Bangladesh's most ambitious policy response to the permanence problem and is now the clearest case study in why policy responses sized to political optics fail to scale. The island was built up from a sediment platform in the Meghna estuary by the Bangladesh Navy. The original Tk 23.12 billion budget approved by ECNEC was equivalent to approximately two hundred and seventy million dollars at the time, with later supplementary allocations bringing total reported public investment to roughly three hundred million dollars. The original capacity target was one hundred thousand persons. Construction was largely complete by mid-2020. The first relocation of three hundred persons occurred in December 2020. The relocation programme reached approximately twenty-eight thousand by mid-2021, paused under international pressure, resumed under a memorandum with UNHCR in October 2021, and continued at a reduced pace.

As of the most recent UNHCR Bhasan Char monitoring updates through 2024, the relocated population was approximately thirty-five thousand. The capacity remains one hundred thousand. The shortfall is two thirds, more than five years after the first relocation began. The Rohingya themselves resist relocation because of restricted movement, limited livelihood opportunities, the documented vulnerability of the platform to cyclone inundation despite the Navy-built embankments, and the simple fact that the social networks that sustain camp life are in Ukhia and Teknaf, not on a remote island.

The Bhasan Char number that matters is not thirty-five thousand. It is the implied policy claim. To meaningfully reduce the Cox's Bazar concentration, Bangladesh would need to relocate approximately two hundred thousand persons over five years. At the current pace and political acceptability, this is not happening. Bhasan Char will likely stabilize at fifty to seventy thousand, with the rest of its capacity unused. The roughly three hundred million dollars of construction will yield housing for roughly half its design population. By any reasonable accounting, this was a partial failure as a containment strategy and a complete failure as a policy signal that Bangladesh had a viable alternative to the Cox's Bazar concentration.

The lesson Bhasan Char teaches Bangladesh's policy apparatus is that physical infrastructure is the easy part of a relocation programme. The hard part is the legal, economic, and social framework that makes a population move and stay. Bangladesh built the infrastructure and avoided the framework. The framework requires accepting the population as durable, providing rights to work, education, and movement, and creating an economic basis for the relocation site that does not depend on indefinite humanitarian subsidy. None of this was done. The result is that two thirds of the camps Bangladesh built sit empty, while the camps that were supposed to be drained remain at full pressure.

What Bangladesh has not addressed

Bangladesh has done many things well in the Rohingya response and the international community has acknowledged them. The country opened its border in 2017 when many states would not have. It has provided physical security, against substantial pressure, to a population it has no treaty obligation to protect, given that Bangladesh is not a party to the 1951 Refugee Convention or its 1967 Protocol. It has cooperated with UNHCR and IOM on registration, biometrics, and protection. The food and shelter operation in late 2017 and early 2018 was implemented at a logistical scale that few governments could have managed.

What Bangladesh has not done is build the policy and legal framework appropriate to a million-person stateless population whose presence has become permanent. The current framework is governed by the Foreign Donations Regulation Act, the Rohingya specific NGO Affairs Bureau guidelines, the Joint Response Plan, and a series of administrative orders that govern movement, employment, marriage, education, and registration. Each instrument was designed for a temporary population. The instruments are now ten years old and the population has crossed one million.

The specific items the framework does not address are:

Right to work. The Rohingya are formally prohibited from formal employment in Bangladesh. They participate in the camp economy and the host community informal economy through arrangements that have no legal recognition. This pushes wages below the cost of subsistence, exposes Rohingya workers to exploitation, and removes the most natural mechanism by which a displaced population reduces its dependence on aid. International experience from comparable cases is unambiguous: a population that cannot work formally and cannot leave is more expensive to host, not less. Uganda's approach to Congolese and South Sudanese refugees, which provides a right to work and freedom of movement, results in lower per-capita aid costs and higher self-reliance scores than Bangladesh's containment model.

Right to documented education. The Myanmar curriculum was introduced in camp learning centres from 2020 onward, supplementing the previous informal arrangements. The certificates issued by these centres are not recognized by any Bangladeshi institution. A Rohingya student who completes the equivalent of secondary school in the camp cannot enrol in any tertiary institution in Bangladesh, cannot sit for any standard examination, and cannot translate the credential into any labour market value either inside or outside the camp. The 2024 cohort of Rohingya children who would be sitting for SSC equivalents if they were in any other school system in the country will turn eighteen in 2026 with no recognized credentials at all.

Civil registration. The births inside the camps since 2017 are not registered with any Bangladeshi civil authority. They are recorded by UNHCR for protection purposes and by Bangladeshi camp administration for population management. They are not on the Bangladesh national identification system, they are not in any civil register, and they have no documented nationality. UNHCR's 2024 statement on the Rohingya described them as the largest stateless population in the world. A growing share of that population has been stateless since birth.

Movement out of the camps. Movement is restricted by checkpoint, by curfew, and by the practical absence of documents that would permit formal travel. The result is a generation that has lived its entire life within twenty kilometres of the Bangladesh-Myanmar border. Whatever the security justifications for movement restriction in 2017 and 2018, they have weakened over time, while the developmental costs have grown. There is no policy review that has tested whether a graduated movement framework could be safely implemented.

Pathway to legal status. Bangladesh's position has consistently been that legal status is incompatible with eventual return and would create a moral hazard for further influxes. This position was defensible when return was a near-term policy goal. It is no longer defensible when return is at least a generation away. Other states that have hosted long-tenured stateless populations have addressed the legal status problem in graduated ways: Tanzania granted citizenship to the long-tenured Burundian caseload in 2010, Uganda's prima facie recognition framework provides effective legal status without citizenship, and Kenya is moving toward graduated integration of long-tenured Somalis. None of these models is directly transplantable. All of them illustrate that the choice is not binary between containment and citizenship, and that the cost of avoiding any framework increases over time.

What changes if the framework changes

The cost of permanence to Bangladesh is not the cost of food, shelter, and water for one million people. That cost is largely covered by the international community, and the gap between the appeal and the funding is the immediate operational problem that brings food insecurity to the camps. The cost of permanence to Bangladesh is the cost of running a containment regime over a generation, with the wage suppression, environmental degradation, and host-community alienation that come with it.

A policy framework that accepts permanence and adapts to it would have several features. None of these is novel. All of them have been tested elsewhere. Bangladesh's reluctance is political, not analytical.

A graduated right to work, beginning with categories of work that do not directly compete with the Bangladeshi labour market and that contribute to either camp self-sufficiency or host community development. Camp-internal economic activity, agricultural work in approved zones, and skilled trades for which there is documented Bangladeshi labour shortage are all candidates. Even a partial right to work would shift the per-capita cost of the operation downward and reduce the wage suppression in Ukhia and Teknaf by formalizing the labour that is already happening informally.

A documented education pathway, beginning with the recognition of Myanmar curriculum certificates within Bangladesh, then progressing to vocational and tertiary access through quotas. The 2024 secondary cohort will turn eighteen in 2026 and the question of what they do for the rest of their working lives, in or out of the camps, is the question Bangladesh's policy framework refuses to answer.

A statelessness register, separate from the Bangladesh national identification system, that records births, marriages, and deaths in the camps under Bangladeshi civil authority and produces documents that have at least administrative value. This does not require any change to the Bangladesh Citizenship Act and does not prejudice the future status of the population. It would replace UNHCR-issued documentation with Bangladeshi documentation for civil purposes within the country.

A graduated movement framework, beginning with documented daily movement for work and education within Cox's Bazar district and progressing to district-level movement under the same documentation framework that applies to other restricted populations. The current binary of containment or no constraint could be replaced with a system of registered movement.

A Bhasan Char rationalization, accepting that the relocation programme as originally framed has not worked, repurposing the unused capacity for productive use, and retaining the relocated population on terms that match the rights framework that applies in Cox's Bazar. The current ambiguity, in which Bhasan Char is neither a successful relocation site nor a closed programme, serves no one.

A bilateral and multilateral funding architecture that recognizes the Joint Response Plan as a component of long-term financing rather than an annual humanitarian appeal. The World Bank's IDA-19 and IDA-20 sub-windows for refugees and host communities are the existing instrument for this transition. Bangladesh has accessed these windows at modest scale. A larger commitment, with co-financing from the EU, the Asian Development Bank, and the OIC member contributions that have not historically materialized at scale, could replace the eroding annual humanitarian financing with predictable multi-year development financing.

None of these changes is cheap. Each carries political costs that the current Bangladeshi political environment will be reluctant to absorb. But the alternative, which is the continuation of the current framework through 2030 and beyond, is more expensive in every accounting that does not show up in the national budget.

The 2026 window

Prime Minister Tarique Rahman's BNP government, sworn in on 17 February 2026, faces this question in its first policy year. Foreign Minister Dr. Khalilur Rahman has stated publicly that the government will pursue repatriation drawing on past BNP-era successes from 1978 and 1992, with Bangladesh having shared biometric data on 829,000 Rohingya with Myanmar for processing. The stated position is that safe, voluntary, and dignified return is the only sustainable solution. The operational reality, as the same Foreign Minister has acknowledged in parliament, is that active conflict in Myanmar means no movement can begin in the near term.

The 2026 window is therefore dual: the BNP government has both a diplomatic mandate to pursue return and a governance obligation to manage a million-person population that will be in Bangladesh for at least the duration of the current Myanmar conflict. The framework reset does not require abandoning repatriation as a goal. It requires administrative orders, NGO Affairs Bureau policy revisions, and a coordinated communication that prepares the host community and the national audience for a shift in framing from temporary containment to managed coexistence pending return.

If the reset does not happen in 2026, the most likely trajectory is the indefinite continuation of the current framework, with the food ration somewhere between six and ten dollars depending on global humanitarian financing, with the host community wage and rent dynamics continuing, and with a generation of stateless children turning twenty without documents, education, or work.

There is an alternative framing of this brief, in which the question is humanitarian rather than fiscal, and the moral case for action is the case for action. That framing has been made many times by many people more qualified to make it than this brief. We have used the fiscal and demographic accounting because that is the framing the policy environment in Bangladesh in 2026 is more likely to respond to. The accounting is consistent with the moral case. They reach the same conclusion through different reasoning.

The local environment

The Cox's Bazar climate data does not show a sharp discontinuity at 2017, which is itself worth noting. Whatever the local environmental degradation in the camp footprint, the regional climate signal is dominated by the slower trend that affects the entire coastal belt. The annual mean temperature at Cox's Bazar has risen from approximately 25.6 degrees in the early 1980s to approximately 26.5 degrees in the early 2020s, a trend consistent with regional warming. The 2017-2024 period sits within the warming trend, not outside it.

The implication is that the climate adaptation challenge for Cox's Bazar district as a whole, which would have existed without the Rohingya influx, is now layered on a population concentration that has reduced the district's adaptive capacity. The Bangladesh climate adaptation budget for Cox's Bazar through the 2030 Bangladesh Delta Plan does not specifically allocate for Rohingya-related demand on water, sanitation, or coastal infrastructure. The climate financing architecture and the Rohingya financing architecture have been kept separate. The populations they serve are not separate.

What this brief does not address

A complete policy review of the Rohingya question would address several items this brief has set aside.

The criminal justice and security situation inside the camps has deteriorated since 2022. The Arakan Rohingya Salvation Army and the Rohingya Solidarity Organization are both active, with documented internal violence and external recruitment. The Bangladesh Armed Police Battalion deployed to the camps from 2020 onward has been a partial response and has been credibly accused by Human Rights Watch of its own abuses. A serious treatment of the security situation requires a brief of its own.

The trafficking and irregular maritime departure question, including the resumption of dangerous boat journeys from Bangladesh and Indonesia in 2022 and 2023, is a regional problem that the Bali Process has not solved and that requires its own treatment.

The role of OIC member states, which have provided rhetorical support for the Rohingya cause but limited financial support for the Joint Response Plan, deserves a separate accounting.

The question of accountability for the 2017 atrocities, including the Gambia case at the ICJ and the universal jurisdiction cases under way in Argentina and elsewhere, is a parallel track that this brief has touched only briefly.

Sources and data lineage

The refugee population trajectory chart is built directly from the World Bank's reproduction of the UNHCR Population Statistics Database, indicator SM.POP.RHCR.EA, refugees by country of asylum. The series is maintained by the World Development Indicators team and is the most widely cited single time series for refugee stocks in academic and policy work. The series is internally consistent across the 1988-2024 period covered.

The Joint Response Plan funding chart is built from the public summaries of the JRP issued annually by UNHCR, IOM, and the UN Resident Coordinator's office in Dhaka. The 2018-2024 figures are from final-year reports. The 2025 figure is from the mid-year review released in August 2025 and is provisional.

The WFP ration trajectory is built from WFP press releases, WFP Country Strategic Plan annual reports for Bangladesh, and WFP Bureau of Asia and the Pacific situation updates. The April 2025 cut to six dollars is documented in WFP's public statement of 18 April 2025 and was confirmed by donor briefings in Cox's Bazar that month.

The Cox's Bazar climate trend is built from NASA POWER 2-metre air temperature reanalysis data for the Cox's Bazar grid cell, aggregated from monthly to annual values. NASA POWER is a satellite-and-model reanalysis product and the values are not station observations. The trend is consistent with Bangladesh Meteorological Department station data for Cox's Bazar where the records overlap.

Host community impact figures are drawn from the World Bank Cox's Bazar Panel Survey waves of 2019 and 2022, the BIDS host community labour market studies, IFPRI's wage and food price studies, and the FAO/UN-Habitat/DLR forest cover analyses cited in the text. Specific numerical claims in this brief have been chosen to fall within the consensus range across these sources rather than at the extreme of any one of them.

The 1992 caseload precedent and the historical UNHCR-monitored returns of the late 1990s are documented in UNHCR's archived Operations Bangladesh files and in the academic literature on the first Rohingya exodus, particularly Brad Adams' 2000 work for Refugees International and the contemporaneous reporting in the Asian Survey academic journal.

The legal framework references are to the Foreign Donations Regulation Act 2016, the NGO Affairs Bureau guidelines for Rohingya response 2017 and updates, the November 2017 Bangladesh-Myanmar bilateral arrangement on the return of displaced persons, and the October 2021 UNHCR-Bangladesh memorandum of understanding on Bhasan Char.

A Bangla-language opinion piece on the same material is available at রোহিঙ্গা: প্রত্যাবাসনের মৃত্যু, হিসাবের জন্ম.

Sources

• UNHCR Bangladesh Operational Data Portal (refugee stock, protection data): https://data.unhcr.org/en/country/bgd
• World Bank WDI SM.POP.RHCR.EA (refugees by country of asylum, 1988-2024): https://data.worldbank.org/indicator/SM.POP.RHCR.EA?locations=BD
• UNHCR/IOM Joint Response Plan 2025-2026 ($934.5M biennial, Jan 2025-Dec 2026): https://reporting.unhcr.org/operational/operations/bangladesh
• WFP Bangladesh Country Strategic Plan and ration cut statement (April 2025): https://www.wfp.org/countries/bangladesh
• RRRC (Refugee Relief and Repatriation Commissioner) Bangladesh: https://rrrc.org.bd
• Cox's Bazar Nutrition Surveillance Round 2, September 2023 (GAM 15.1% threshold): https://www.unhcr.org/us/where-we-work/countries/bangladesh
• Bangladesh Foreign Ministry statement on 829,000 biometric data shared with Myanmar (Dr. Khalilur Rahman, Parliament, 2026): https://www.bssnews.net/js-session/382270
• IOM Bhasan Char monitoring updates (2024): https://www.iom.int/countries/bangladesh
• World Bank Cox's Bazar Panel Survey 2019 and 2022: https://microdata.worldbank.org/index.php/catalog/3805