The Water Compact

The problem

In 1922, seven states sat down at Bishop's Lodge in New Mexico and divided the Colorado River. They allocated 16.5 million acre-feet a year among the upper basin, the lower basin, and later Mexico.

The river has never reliably carried it. Tree-ring reconstructions show the negotiators had the misfortune to measure during one of the wettest multi-decade stretches in roughly a thousand years. Actual natural flow has averaged closer to 12 to 13 million acre-feet, and the twenty-first century average is lower still.

So the foundational document of the American West over-allocates the river by something like a quarter, and has done so, on paper, for over a century. Forty million people and a large share of the nation's winter produce sit downstream of that arithmetic. Lake Mead and Lake Powell have spent the last two decades demonstrating what happens when you withdraw more than arrives: the reservoirs are the buffer, and the buffer is what has been consumed.

The 2007 interim operating guidelines that have governed shortage-sharing on the river expire in 2026. The successor regime is being negotiated now. This is the rarest thing in water policy, a scheduled moment when the rules are actually open.

Underground it is worse, because underground nobody is even counting. The High Plains Aquifer, the Ogallala, underlies eight states and supports roughly $20 billion of agricultural output. Parts of western Kansas and the Texas Panhandle have already lost more than half their saturated thickness, and in some districts the remaining economic life is measured in decades, not centuries. In much of Texas, groundwater is still governed by the rule of capture: the water under your land is yours if you can pump it, and your neighbor's dry well is your neighbor's problem.

You cannot manage what you do not measure, and American water is, to a startling degree, unmeasured.

The proposal

Meter the water, put every right on a public ledger, let rights trade inside a basin, and buy back the overdraft from willing sellers rather than litigating it out of them.

How it would work

1. Measure first. Federal cost-share for telemetered metering of every major surface diversion and every high-capacity well in the Colorado and High Plains basins. Data published, at the point of diversion, with a lag short enough to matter. Everything below this depends on it, and none of it is possible without it.

1. A public register of rights. Priority date, quantity, place and purpose of use, in one machine-readable ledger per basin. Western water rights currently live in county courthouses, state engineer files, and the memory of specific lawyers. A right nobody can look up cannot be traded, valued, or retired.

1. Trading inside the basin, with the river protected. Rights become transferable, permanently or by annual lease, subject to a no-injury test for downstream users and a hard cap on out-of-basin export. This is not new. Australia's Murray-Darling market moves water worth billions a year, and California's SGMA has already forced groundwater accounting on a state that resisted it for a century.

1. Buy the overdraft down. A federal facility purchases and permanently retires consumptive use from willing sellers at auction, targeting the gap between allocation and hydrology. Reverse auction, lowest bid first, funded on a multi-year appropriation so it can act countercyclically when farm economics are weak and rights are cheap.

1. Pay the community, not only the landowner. A retired water right takes a farm's water and a town's tax base with it, and the buyout as usually designed pays the first and ignores the second. A share of every purchase goes to the county in a permanent local endowment, structured like the state funds of № 009. The town that gives up its water keeps an income from having done so.

1. Nothing is taken. Every mechanism here is voluntary and compensated. Water rights are property, the takings jurisprudence around them is serious, and a proposal that pretends otherwise is a proposal that spends fifteen years in court.

The numbers

The Colorado's structural deficit is commonly put at 2 to 4 million acre-feet a year against current allocations.

Agriculture consumes roughly 70 to 80 percent of the river's water, and forage crops, alfalfa and grass hay, are the single largest use. This is the central fact of Colorado River policy and it is why the conversation is not really about lawns or golf courses.

Recent voluntary conservation agreements in the lower basin have paid on the order of $300 to $500 per acre-foot for temporary reductions. Permanent retirement costs more; call it $2,000 to $4,000 per acre-foot of consumptive use, varying enormously by seniority and district.

Retiring 1.5 million acre-feet permanently at $3,000 is roughly $4.5 billion, one time.

Set that against the cost of not doing it. Lake Mead below minimum power pool ends generation at Hoover Dam. Below dead pool, water stops passing the dam at all, for Las Vegas, Phoenix, Los Angeles, San Diego and the Imperial Valley. There is no honest cost estimate for that outcome because it has no modern precedent, which is itself the argument.

$4.5 billion is roughly four days of federal borrowing. The West has spent longer than that arguing about a single diversion.

The honest objections

"This is a wealth transfer to whoever holds the most senior rights, and those are often the largest operations." Substantially true. Seniority in prior appropriation reflects who showed up first, not who farms best or needs it most, and buying out senior rights pays the historical accident handsomely. We accept this because the alternative is worse: a system that reallocates by curtailment produces a decade of litigation and no water, while a system that reallocates by purchase produces water. Paying an unfair distribution to move quickly is a real cost and we are counting it as one.

"Markets will drain rural communities to water suburbs and data centers." The pattern has a name, buy-and-dry, and Crowley County, Colorado is what it looks like: rights sold to Front Range cities, farms gone, town hollowed. Item 5 exists for exactly this and we do not claim it fully solves the problem. A county endowment is not a farm economy. The out-of-basin cap in item 3 is the harder constraint, and the honest position is that some agricultural land in the West is going to stop being farmed, because there was never enough water for all of it and the last century of pretending was the anomaly.

"Metering wells is politically impossible in Texas and Kansas." It is extremely difficult, and rule-of-capture states have defeated this repeatedly. The federal government cannot impose it. It can pay for it, which is a different proposition, and it can condition the substantial federal money already flowing to irrigation districts on participation. That is leverage rather than mandate, and it may well be insufficient. If a basin declines, it keeps its rule of capture and its aquifer decline, and neither this proposal nor any other can help it.

"Tribal rights are unquantified and you have written a market on top of an unsettled question." The most legitimate objection to the mechanism. Many tribal reserved rights under Winters remain unquantified, they are senior, and a trading system built before quantification would allocate around claims that legally come first. The sequencing has to be explicit: settlement and quantification funding precedes market opening in any basin, not the reverse. A market that quietly prices around unsettled tribal claims would be a second dispossession using a modern instrument.

"The 1922 Compact cannot be reopened." Probably right, and this proposal deliberately does not try. Everything above operates on the use of water rather than the interstate apportionment of it. Reopening the Compact invites seven state legislatures, Mexico, thirty tribes and the Congress into a single negotiation, and that negotiation has no plausible end. Work under the Compact, not on it.

Sources

• Colorado River Compact of 1922; allocation of 16.5 MAF against a twentieth-century natural flow averaging roughly 12–13 MAF; tree-ring reconstructions of long-run flow
• 2007 Interim Guidelines for Lower Basin shortages and coordinated operation of Lakes Powell and Mead, expiring 2026 (usbr.gov)
• High Plains (Ogallala) Aquifer saturated-thickness change; US Geological Survey groundwater monitoring (usgs.gov)
• Agricultural share of Colorado River consumptive use, with forage crops the largest single category
• Australia's Murray-Darling Basin water market; California Sustainable Groundwater Management Act (2014), sustainability required by 2040
• Winters v. United States (1908), on federal reserved water rights for tribal lands
• Proposals № 009 (The Permanent Fund Compact, the model for local endowments); № 024 (The Grid Dividend); № 036 (The Rural Compact)