When a climate disaster ravages remittances and dreams

Two jobs at once 

Drive on any road through Nepal’s middle hills and one can read the migration history by the rooflines. The pillar house — a reinforced concrete frame, rendered and painted, standing where a mud-and-stone house used to be — is the visible proof that someone in the family went to the Gulf, or to Malaysia, and that the money was sent back. Often the frame is unfinished. One storey is complete and lived in; above it, lengths of steel reinforcement bar rise out of the columns into open air, rusting, waiting for a second floor. The rebar is not an oversight; rather, it is a plan. It says another contract will be signed, another two or three years will be spent abroad, and the next instalment will pay for the next floor.  

This is the loop that many migrant households retrace. Migration generates remittances; remittances become houses, land and shops; those assets sit where floods and earthquakes find them; and their loss can push a household to send someone abroad again. The usual debate asks whether migration builds resilience or deepens vulnerability. The Nuwakot case suggests it does both at once. It builds the buffers that carry households through a shock, and it concentrates their savings in immovable, exposed form. Which effect wins depends on conditions that can be named, many of them outside the household’s control.  

A warming climate is fast becoming the most important of those conditions. Across the Himalayas, rising temperatures are thinning glaciers, thawing the frozen ground that holds high rock walls together, and turning snow into rain. The floods and landslides that follow come down to the river terraces and highway bazaars where migration money is spent. The climate vulnerability and risk assessment already counts floods and landslides among the forces pushing young people out of farming and into work abroad - Opens in new tab. A climate disaster can therefore act on both ends of the loop at once: destroying what migration built, and sending someone out to build it again. 

In this household, the next floor had been bought. The plan announced by the rebar had been carried out, and the migration cycle behind it had been closed intentionally: the couple came home ‘for good’ about six years before the flood.  

More than a decade of work abroad had made this family secure and at the same time tied most of its savings to a house, two small businesses and a plot of land in a river valley. A flood took that security away in minutes. No one recorded the loss as a migration loss, because damage surveys count buildings and migration statistics count departures, and neither ask how a house was financed. And a family that had finished with migration may now be pushed back towards it. 

That is what makes the case analytically useful rather than only affecting. It is not the story of a household that still has someone abroad. It is the story of one that had finished — and of what a disaster does to a migration that was meant to be over.  

The Flood

The disaster began at about 8:40 that morning, when an estimated square kilometre of rock and glacier ice - Opens in new tab detached from a north-facing slope above the Lhende Khola river, on the flank of Langtang Lirung mountain peak, and fell some 1,200 metres - Opens in new tab into the valley of the stream that marks the border with Tibet, China. Scientists believe - Opens in new tab the collapse dammed the Lhende Khola and that the water building up behind it then broke through into the Bhotekoshi and the Trishuli river system; whether the river was fully or only partly blocked, and for how long, is still being worked ou - Opens in new tabt - Opens in new tab. The channel torn down by the surge was still lined with loose sediment from the glacial lake outburst that had come down the same river on 8 July 2025, which probably made it larger - Opens in new tab. The hydrological station at Syabrubesi stopped transmitting at about 8:50 a.m., its last reading 1.62 metres; by 9:15 the Department of Hydrology and Meteorology had confirmed a major flood wave and issued warnings downstream, and some 679,000 warning text messages went out. None of them reached Timure in time: the alerts were sent at least fifteen minutes after the bazaar was already under water. For Narayan’s family further downstream, as he tells it, the news and the river arrived almost together. By 17 September, authorities had recovered 1,410 bodies and were counting 6,150 people as missing - Opens in new tab.  

The districts hit hardest were Rasuwa and Nuwakot, with Dhading downstream. All three were among the fourteen districts that the Nepali gGovernment classified as severely affected - Opens in new tabby the 2015 earthquake. Much of what the water took in those valleys had been built, or rebuilt, in the eleven years since — reconstruction that was partly financed - Opens in new tab by money earned abroad.  

That is the starting point for three questions worth investigating. When a flood destroys a house built with wages from Kuwait, has it destroyed the migration capital? If it has, is the rebar still a plan — does the household send someone back out to build again? And what happens when the household had already decided that nobody would go again? 

Migration as accumulation 

Nepal converts labour mobility into household income at a scale matched by few economies. The central bank’s annual review - Opens in new tab of the fiscal year that ended mid-July 2026 was released on 26 August — the morning of the flood. It reported remittance of NPR 2.36 trillion (~USD 16.19 billion), up 37% from the previous year’s. The 2021 census - Opens in new tab counted 2.19 million people absent abroad, drawn from 23.4 percent of all households.  

What matters analytically is not the aggregate but the conversion. Remittances arrive as a flow and are held as a stock. Some of that stock is liquid — cash, savings, gold accessories — and some is fixed: a house, a plot of land, a shop, a herd. The distinction is not just accounting. Liquid assets survive an earthquake in a way that a house does not. The rapid assessment carried out in four crisis-hit districts - Opens in new tab weeks after the 2015 earthquake found exactly that: where migrant households reported an advantage, it lay in having cash, food stocks and jewellery to hand, not in having stronger buildings.  

Fixed assets nonetheless dominate the vernacular of Nepali migration. A house in the village is what migration is understood to be for. Analysis of the 2010/11 living standards survey - Opens in new tab for the central hill and mountain region — which contains twelve of the fourteen districts later classified as severely earthquake-affected — found households with a member abroad more likely to sit in the richest consumption quantile and holding substantially more valuable dwellings than average. Households with an internal migrant were poorer than average on the same measures, a distinction that we will return to later. Both comparisons are cross-sectional and the region includes the Kathmandu Valley, which inflates the external-migrant figures; they establish association, not a causal effect of migration on wealth. The Valley’s three districts, Kathmandu, Lalitpur and Bhaktapur are among the most affluent in the country, and its households with a member abroad lift the averages above what migrant households in hill districts such as Rasuwa or Nuwakot would show. 

Narayan and Laxmi’s account is an unusually explicit version of that vernacular, with one difference: the accumulation had an end date. “We were confident that now we could make our living — in fact, live a decent life — in Nepal,” Narayan says, “in our hometown, Nuwakot.” It was also a joint project, and not only a financial one. Laxmi trained as a beautician before she migrated, practised the trade in Kuwait, and on their return made it her own business. Migration capital here includes a skill acquired at home, deepened abroad and converted into a local enterprise: an asset no assessment will count, and one a flood can destroy without touching the skill itself, by taking the premises and the local market in which it earned. 

Returning for good  

Research on return migration has long distinguished returnees who come home ready from those who are sent or pushed home. In Cassarino’s influential formulation - Opens in new tab, the outcome of return depends on a returnee’s preparedness — willingness and readiness to return — and on the resources, tangible and intangible, that he or she can mobilise. By that yardstick, this household was exceptionally well prepared. The return was by choice, with savings, a marketable skill and, in Narayan’s words, “established connections in the potential market” of Kathmandu, where he and several relatives first intended to sell goods brought through the Chinese border crossing. They went north to set up the trade, and he changed his mind.  

Once we were there, picturesque Timure prompted me to start a hospitality business, a hotel. The place is a transit point to China and to one of the most sacred pilgrimages, Kailash Mansarovar in China. 

The logic was sound. The Rasuwagadi-Kerung crossing is one of the routes Nepal offers to Kailash Mansarovar, a pilgrimage undertaken by hundreds of thousands annually, and the trade and pilgrim traffic through Timure was the reason anyone built there. 

 It was also a business whose revenue depended entirely on the border staying open. The Kailash - Opens in new tabyatra - Opens in new tab through Nepal was halted in 2020 after China closed its border points - Opens in new tab, and Kerung reopened fully for two-way trade and the movement of people only from April 2023.  

The hospitality business was smooth, but COVID appeared out of nowhere and forced us to shut it down. The house owner took it from us and started running it themselves. I was in regular touch with them. The business had picked up; I was happy for them. But the flood didn’t spare them. The whole of Timure was swept away. The owners, who were attending to their customers, did not survive.  

I used to lament that I had to leave that place. Now, having survived this devastation, I think: what if COVID hadn’t come, what if I hadn’t had to sell my hotel? We would be among the dead as well.  

The pandemic cost this household its first reintegration investment. The family moved down to Nuwakot, their hometown, and started again.  

We built a new house, and my wife ran a beauty parlour. I set up a poultry farm. We were doing decently.  

By the most widely used standard, that is what a successful return looks like. The International Organization for Migration (IOM) treats reintegration as sustainable - Opens in new tab when returnees have reached economic self-sufficiency, social stability in their communities, and enough psychosocial wellbeing to withstand the pressures that drive migration, so that any further migration becomes ‘a matter of choice, rather than necessity’. Six years after coming home, and after one venture had already been lost, this household met that standard on its own.  

For all their years in the Gulf, the couple does not have an active pension that pays them regularly. Kuwait, like most Gulf States, excludes migrant workers from its national social insurance - Opens in new tab. - Opens in new tab Narayan and Laxmi received the one entitlement granted to most migrants: an end-of-service lump sum. At home, Nepal began enrolling its workers abroad in its contributory Social Security Fund only in March 2023, - Opens in new tab about three years after the couple came home. The Foreign Employment Welfare Fund, financed by a levy on departing workers, compensates families for death, injury or serious illness during a contract or within a year of its end - Opens in new tab, but not for savings lost years later to a river.  The house, the parlour and the poultry farm were more than just investments; they were the family’s pension. 

What counts as migration capital 

Migration-generated capital is the portion of household wealth accumulated directly or indirectly through migration income — which requires knowing how a given house, plot or shop was actually paid for, and in what proportion. Almost no post-disaster assessment collects that information. Nepal’s 2015 needs assessment - Opens in new tab, comprehensive as it was, contains no disaggregation of damage by household migration status anywhere in its 23 sectors.  

The 2026 flood has exposed the same blind spot. A preliminary government estimate puts its damage and loss at NPR 408 billion (~USD 2.6 billion), more than two thirds of it borne by private households and businesses - Opens in new tab. H - Opens in new tabow much of that private wealth was earned abroad, nobody has asked. The indirect evidence, however, points one way. After the 2015 earthquake, a study of affected households - Opens in new tab found remittances to be the most important way families coped, and the only one that measurably helped them recover their livelihoods; the aforementioned rapid assessment in four crisis-hit districts - Opens in new tab found that women there had mostly asked their husbands abroad not to come home, but to stay on and send more money for rebuilding. In the weeks after this flood, migrants described - Opens in new tab the family homes they built from Gulf wages and lost in seconds, and labour rights investigators warned that workers who had lost their homes now needed their jobs abroad more than ever. 

The assets built around that migration capital sat where exposure and value coincided: the roadside, the highway, the border. The International Centre for Integrated Mountain Development explains why Timure was among the hardest hit: it occupies the first flat stretch of riverbed below the Lhende’s plunge, where a debris flow drops its load and where the bridge, customs yard, dry port, and hydropower station had been built. The same valley had flooded thirteen months earlier - Opens in new tab, closing the crossing for six months. Across the corridor, hotels and restaurants were the largest group of damaged businesses: 1,259 of nearly 4,800 - Opens in new tab.  

The Nuwakot house failed differently: its frame held but its site did not. Construction type did not mitigate the 2026 damage — the assessment’s - Opens in new tab count includes 2,036 reinforced-concrete homes — because a debris flood tests a location before it tests a frame.  

We hope to rebuild; we hope the government will assist us. But the place where we built the house is uninhabitable, because the river has changed its course. Ours is almost a kilometre away, and yet it has been turned into a riverbank. 

The minister for infrastructure development - Opens in new tab has described the same thing from the road: from downstream, the river has shifted towards the hillsides. The state’s answer is relocation — integrated settlements built over about two years, a monthly rental allowance of NPR 15,000 in the meantime, and a provincial commitment - Opens in new tab not to rebuild along the Bhotekoshi and Trishuli. Relocation will strand land that is legally owned, economically dead and useless as collateral. It also leaves this family with a house that is standing, buried and lost at once, a case that damage categories will struggle to record.  

Counting what was lost 

Standard damage figures miss most of what migration built. For this household, the losses come in layers: a physical asset buried to its second floor; livelihood capital in the parlour and the poultry farm; the migration capital, a decade of Gulf earnings, that those assets embodied; land made worthless by relocation; and what the literature on loss and damage - Opens in new tab calls “non-economic loss – attachment to a place and a life’s work –  that cannot be bought back.” 

We were rescued and taken to a safe place, but my wife and I go to the house every day. We look at it and think of the past — the hardship we went through, living far from our family, to earn a better future.   

Two further gaps matter for policy. Of the 3,916 people first reported - Opens in new tab missing from 62 districts, 639 were recorded at hydropower projects — listed under the plant that employed them rather than under a home address, drawn from across the country to build and run the plants (one group of survivors had come from Bardiya, in the western Tarai - Opens in new tab), and they come from the kind of household the living standards survey found to be poorer than average. Migration data counts departures, damage assessments count structures, but neither asks how a house was paid for.