The United Nations University Institute for Water, Environment and Health published Global Water Bankruptcy in 2026, and the shift in language is deliberate. A crisis is a shock you recover from. Bankruptcy is what happens when the recovery stops arriving. Societies have been drawing down not only the annual income of rainfall and river floww but the savings held in aquifers, glaciers, wetlands and soils — and in many basins those savings are simply gone.
The report's power is in its honesty about permanence. Around 70% of the world's major aquifers are in long-term decline, many of them irreversibly, because once an aquifer compacts it loses the pore space that made it a reservoir in the first place. More than 6 million square kilometres of land — close to 5% of the planet's land surface — have already subsided from over-pumping. Over 30% of glacier mass has been lost since 1970. Roughly 410 million hectares of wetland, an area approaching the whole European Union, have been drained in five decades.
None of that comes back. Compacted aquifers do not rebound, subsided deltas do not rise, and drained wetlands are not restored inside a planning horizon.
And agrriculture sits at the centre of the ledger: about 70% of global freshwater withdrawals go to farming, and roughly half the world's food is produced in regions where total water storage is already declining or unstable.
There is a second kind of water in the report that rarely makes headlines. Hydrologists call it green water — the moisture held in soil, available to plants, never metered or piped or traded. It is the largest freshwater store on land after ice, and the report concludes its planetary boundary has already been crossed, alongside those for climate and land systems.
This is where farming stops being a bystander. More than half of the world's agricultural land is now moderately or severely degraded, and degraded soil holds less water. When soil is compacted, eroded or crusted over, its capacity to absorb rain can fall by as much as 90%. The consequence is the drought–flood paradox that has become ordinary across drylands and increasingly familiar in Central Europe: rain arrives, runs straight off the surface as flash flooding, and the root zone underneath stays dry anyway. The same field floods and droughts in the same season.
A soil that cannot absorb rain has effectively been foreclosed on, whatever the rain gauge says.
This is the part we can measure, and have been measuring since 2025 across paired regenerative and conventional fields in Czechia and Slovakia — neighbouring farms, comparable crops, comparable weather.
In the 2025 campaign, water infiltration on regenerative cropland was significantly faster than on the conventional controls (Mann–Whitney U, p = 0.003). The individual pairs show how large the gap can get. On one arable pair, water entered the regenerative soil at 501 mm/h against 24 mm/h next door. On another, 474 against 62. On a paddock-grazed pasture, 462 against 46.
The mechanism is biological, not mechanical. Earthworm abundance was the single largest response in the 2025 dataset, significantly higher on regenerative pasture — and on one pair, eight worms per block on the regenerative side against none at all on the conventional. Worm burrows and undisturbed root channels are what build the stable vertical macropores through which water travels. Tillage severs them; bare surface soil crusts and seals. Lower bulk density on regenerative pasture (1.17 against 1.34 g/cm³) tells the same story from the other direction: pore volume built by roots and fauna rather than by a machine, and therefore durable rather than temporary.
Across the wider portfolio of 90 farms, bare soil fell by about 11% between the pre-baseline year and 2025, with 74 of those 90 farms carrying more cover than before. Satellite land-surface temperature ran on average 0.5 °C cooler on regenerative fields every month from May to October. Less exposed ground, more shade, more transpiration, less water lost to the air before a root ever reaches it.
The most useful idea in the UN report, for anyone working in soil, is the distinction it draws between water as a product and water as a process. For centuries our laws and institutions have managed water as a measurable good — cubic metres, permits, pipes, tariffs. Far less attention has gone to the natural systems that produce it: soils, wetlands, forests, aquifers, snowpack. Protecting the product while the process is dismantled is not a strategy.
Which is why the report's own prescription for agriculture reads the way it does. Among its recommendations, it calls for on-farm practices that improve soil structure and organic matter — conservation tillage, cover crops, agroforestry, controlled grazing — precisely because they increase infiltration, reduce runoff and rebuild soil moisture storage.
That is a description of regenerative agriculture, written by a UN institute, in a report about irreversible loss. It is the section where the ledger stops closing accounts and opens one.
On our own fields we measured water moving into regenerative soil up to twenty times faster than into the conventionally farmed field next door. It’s all the variables coming together which is rare since inputs and outputs are most of the time uncontrollable.
Our 2026 field campaign did not repeat the infiltration result. Across eight Czech farms, the difference between regenerative and conventional fields was slightly negative and nowhere near statistical significance — −7.6% on cropland, −3.9% on pasture. We are reporting that because it is what we found.
Infiltration is a difficult measurement. It is strongly right-skewed and highly sensitive to how much rain fell in the preceding days and when the reading was taken, which is why one campaign on eight fields settles nothing in either direction. What held steady across both years were the slow accumulative indicators — organic carbon 22% higher on regenerative pasture, nitrogen 13% higher, bulk density and penetration resistance both better — the ones that build the pore structure in the first place. Soil hydrology is a multi-year story measured in multi-year series, and anyone offering you a single season as proof is selling something.
The geography deserves the same precision. The countries at the sharp end of water bankruptcy are not Czechia and Slovakia; they are Iran, Türkiye, India, and the drylands where aquifers have already been spent. Central Europe is not bankrupt. It is the place where the prescription can still be tested before it has to be applied under emergency conditions — and where, on the evidence we have, it works.
Rebuilt soil will not bring back a glacier or refill a compacted aquifer. The UN report is clear that nothing will, and it would be dishonest to imply otherwise.
But soil moisture capacity is the exception in a document full of write-offs — the one water asset that responds to management, on a timescale of years rather than geology, using tools a farmer already owns. Every hectare that starts absorbing its rainfall again is a deposit into the only account still taking them.
Eight co-benefits, one living system, and one credit that measures the tonne of carbon at the centre of it. This is the one the next decade will ask about most.
Sources:
https://collections.unu.edu/eserv/UNU:10445/Global_Water_Bankruptcy_Report__2026_.pdf
