Tunisia is among the world’s leading date exporters by value. The oases that produce those dates sit on confined aquifers that recharge over millennia, if at all. When the crop is exported, the water is not: the commodity and its revenue leave, and the permanent depletion stays behind — priced nowhere on the invoice.
A new preprint from the studio puts a number on that gap. On the most conservative defensible basis, the fossil water consumed to grow Tunisia’s exported dates is worth about half the revenue those exports earn.
1. One dollar of water in every two dollars of dates
Valuing the permanently lost water at what it would cost to replace it — the desalination backstop, the only substitute for the municipal supply the same aquifer ultimately feeds — gives a 2024 debt of $161 million against $305 million of export revenue: 53%.
That number only means something with its range attached. Stack every adverse assumption together and it falls to ~17%. Take the high end of the replacement price and it reaches ~83%. Count irrigation conveyance and percolation losses as permanently gone — which, for a confined fossil reserve, they are — and it passes parity at ~106%.
The headline prices the consumptive quantity — the water biologically and permanently evaporated — not gross extraction. That is the conservative choice, and it understates what the fossil stock actually loses.
2. Why this is not a water footprint
Virtual-water accounting made water in trade visible, but it treats a cubic metre of fast-recharging river water as equivalent to a cubic metre of Pleistocene-age confined groundwater, and it stops at volume.
Three things are different here. The measure isolates the non-renewable fraction. It prices that fraction rather than counting it. And it carries the resulting debt on the traded product, at the commodity level — which is the unit at which trade rules and supply-chain obligations actually operate.
3. What the records already admit
The strongest evidence that extraction is undercounted comes from the administration’s own books, before any modelling enters.
Satellite gravimetry does one job in this argument and no more: GRACE and GRACE-FO confirm, independently of any administrative record, that the basin is losing storage. They are far too coarse to attribute that loss to dates, and the study does not use them for that.
4. The trap: efficiency that isn’t
Reporting the debt per tonne alone would invert the finding. Water per tonne fell over the study period, because yields rose and divided a roughly fixed per-hectare requirement across more output. Read naively, that looks like progress.
Total extraction more than doubled over the same years, and exported tonnage roughly tripled. Export earnings per unit of mined water barely moved across twenty-two years despite revenue nearly tripling — the growth was underwritten by mining far more fossil water, not by mining it more valuably.
5. Where it becomes legible
A commodity-level depletion measure matters because that is the unit emerging supply-chain regimes work in. The EU Corporate Sustainability Due Diligence Directive obliges large in-scope firms to account for adverse environmental impacts across their chains of activity, and the impacts it names include excessive water consumption and degradation that impairs access to safe drinking water.
Two limits keep that modest. The regime is still being built, so the significance is legibility rather than an imminent deadline. And such obligations attach only to EU importers — about 44% of this trade by value in 2017 — while the single largest destination, Morocco, sits outside the regime entirely. The framework’s value is scientific before it is regulatory.
Limitations
This brief summarises a preprint that has not yet undergone peer review.
The replacement cost is an accounting price for what supplying the water will cost once the stock is gone, anchored to the municipal supply the depletion threatens. It is not a market price or a willingness-to-pay measure for irrigation water, and results are reported across the full $0.24–$1.10 per m³ band rather than at a point.
The study does not attribute the observed aquifer depletion to dates via GRACE, does not compute a formal natural-capital set-aside, and does not observe whether the exporting economy reinvests these rents elsewhere — so it makes no claim of a formal Hartwick failure. Export volumes before 2018 are interpolated between six directly reported years; the result holds on those anchor years alone. Which governorate’s dates reached which country is not recoverable from the data, so no origin-by-destination claims are made.
Sources
This brief is derived from the preprint:
Gasmi, T., Guesmi, R., Abdelbari, S., Boulaares, S., Albarghati, S. (2026). The fossil-water debt of agricultural exports: satellite gravimetry and a climate–yield re-coupling index, demonstrated for Tunisian dates. EarthArXiv. https://doi.org/10.31223/X5KV3F
- Data and code: github.com/tanitdata/DatePalm (
ecological_debt/), archived at 10.5281/zenodo.21251191 - Companion study, peer-reviewed, which establishes the aquifer decline this brief builds on: Gasmi et al. (2026), Agricultural Water Management — see Southern Tunisia’s date oases
- Primary sources: ONAGRI (date production by governorate); CRDA Tozeur, Kébili, Gafsa, Gabès (extraction volumes and well records); NASA JPL GRACE/GRACE-FO mascons RL06.3; ERA5-Land (Copernicus); INS DataStore and UN Comtrade (HS 080410) for export volume and value
- AI-native access to the Tunisian portals: the study read agridata.tn and the INS data portal programmatically through the studio’s agridata MCP and INS MCP servers