Groundwater is being extracted faster than an economically optimal rate in three-fifths to four-fifths of the aquifers examined in a new study, according to researchers who modeled the long-term costs of pumping.
The October 2026 working paper by Brian Greaney, Joseph S. Shapiro and Katherine R. H. Wagner compares observed extraction with modeled benchmarks across many of the world’s roughly 3,400 aquifers. It estimates welfare losses in the trillions of dollars in present value, a measure that converts future costs and benefits into their value today.
The figures are model estimates, not a bill already incurred or a count of aquifers proven to be running dry. The authors’ conclusions depend on assumptions about water stocks, the costs of pumping, the value of water in production and how much weight is given to future use.
What is an aquifer?
An aquifer is a body of permeable rock or sediment that stores and transmits groundwater. Wells can draw water from it for irrigation, drinking water or industrial use. Some water is replenished by rainfall and other flows, although recharge varies greatly by location and can take a long time.
When a water table falls, a pump may need to lift water from greater depth. That can increase energy and equipment costs. A farmer’s decision to pump today can also affect the amount available to other users and to the same farm in future years.
Our earlier reporting on groundwater losses in the Colorado River Basin examined physical changes during a particular drought. The new working paper asks a different question: whether the timing of use is economically efficient across many aquifers.
The cost of choosing water today
The researchers use remote sensing and economic data to construct a model for each aquifer. They compare the pattern of extraction implied by their estimates with benchmarks that place different values on future consumption.
An economically optimal path balances the benefit of using water now against the future cost of lower water levels. It does not automatically mean leaving every aquifer at its current level forever. The authors report that a policy holding current fill constant indefinitely can produce less modeled welfare than their optimal path, partly because many stocks are already below the level their model considers efficient.
That conclusion should not be read as a recommendation to drain aquifers. It illustrates that the best modeled policy depends on the starting condition, recharge, future demand and the costs of moving toward a different water level.
Why individual decisions may not add up
Groundwater is often shared by farms, households and businesses. A user who benefits from pumping may not bear all the costs imposed on neighboring wells or later users. Rules governing water rights, electricity subsidies and access to pumps can influence those decisions.
The paper finds larger modeled welfare costs under some policies that guarantee fixed water quantities indefinitely or give no weight to future extraction. Local regulation still has to account for water rights, monitoring costs and the households or businesses that depend on existing supplies.
For businesses with water-intensive operations, the question is where water comes from and whether the same source is shared by suppliers. A food manufacturer may rely on irrigation indirectly through crops purchased from several farms in one basin.
Estimates need local checks
Satellite observations and global models make broad comparisons possible, but they cannot replace detailed measurements of individual wells. Recharge rates, crop choices, electricity prices and legal arrangements differ across regions. Results for an individual aquifer may change as better local information becomes available.
The research is a working paper circulated for discussion. Its monetary estimates describe welfare across time under a particular model, not an immediate loss of cash or an audited market value. For a local business, well records and basin-level assessments remain necessary to judge the risk of future pumping costs.
