In 2026 the Imperial Irrigation District paid its growers $334 to leave an acre-foot of Colorado River water in the river. It charges those same growers $20 to deliver one.
Both numbers are current. Neither is a typo. I spent a while assuming one of them had to be wrong, and the reason they can both be true is not incompetence and it is not greed. It is a ballot measure from 1996.
The twenty dollars is a ceiling, not a price
IID holds rights to roughly 3.1 million acre-feet of Colorado River water a year, about a quarter of what the basin carries in an average year, and it moves that water through some 3,000 miles of canals and drains. Its federal contracts exempt it from paying for the water itself. So the only thing it can lawfully bill a grower for is the carrying.
Twenty dollars is the cost of conveyance. It is not the price of the thing conveyed, because there is no lawful line on the bill for that.
That figure is not an estimate. IID’ Water Rate Schedule No. 1, General Agricultural Service, reads: for all water delivered, $20.00 per acre-foot. The document carries two dates at the foot of the page — board action July 21, 2009, effective January 1, 2011.
Proposition 218 is why. Passed in 1996, it restricts property-related fees to the actual, proportional cost of service delivery. A district that wanted to charge its growers more would have to prove higher service expenses in a formal cost-of-service study, give landowners 45 days’ notice, and then survive a majority-protest process in which the people paying the fee can vote it down in writing.
That is not a rate the board revisits each spring. It is a cap.
And it has held for a long time. The agricultural schedule and the industrial schedule were both adopted by the same board action, on July 21, 2009. The industrial rate took effect that August, the agricultural rate the following January. Neither number has moved since. An entire cycle of drought, shortage declarations, federal renegotiation, and a basin-wide reckoning over the Colorado River has passed over a price set at one meeting seventeen years ago, and the price did not notice.
A price that cannot rise cannot ration
Here is the same water — same river, same canal, same year — priced five different ways depending on who is holding the glass.
| Who | Per acre-foot | Source |
|---|---|---|
| Imperial Valley grower | $20.00 | IID Water Rate Schedule No. 1 |
| Industrial user on the district’s canal system | $85.00 | IID Water Rate Schedule No. 7 |
| What conservation costs IID | $300–$450 | published estimate |
| What IID paid growers in 2026 not to use it | $334 | district conservation program |
| San Diego County Water Authority, wholesale | $700–$1,200 | reported range |
Schedule No. 7 is worth reading closely. It applies to all industrial users who divert from the district’ canals, at $85.00 an acre-foot — except, it says, where there is a signed contract between the water user and the board. That exception is the door every large new project now goes through, and the fees on the other side of it are not $85.
Now read the fourth line against the first. The district pays about sixteen times the delivery price to have the water not delivered.
That is the tell. A price is a job, not a label — it signals, it rations, and it tells a stranger what a thing is worth. I wrote about what a price is actually doing when it rises in about the simplest terms I could manage. Cap it and none of that work gets done. So you buy the behavior you would otherwise have charged for.
Fallowing programs in the Imperial Valley are not primarily a conservation philosophy. They are what is left when the price is fixed by law and the water is still needed somewhere else.
A grower deciding whether to plant alfalfa this year is not comparing $20 against $700. He is comparing $20 against $334, because $334 is the only competing offer the law lets him see.
There is a familiar shape to this. Who stands closest to the cheap tier decides who captures the gap, which is the same distributional question as the Cantillon effect — there, new money reaching some hands before others; here, a fixed price reaching some users and not others. The valley’ water has a front row too.
The best version of the argument for the cap
I want to put the other side at its strongest, because it is considerably better than the arbitrage story makes it sound.
Start with the fact that these water rights are not individually owned. A California Court of Appeal held that IID holds the rights in trust and that individual landowners have a right to water service, not an appurtenant property right in the district’ water. So “let the growers sell it” describes a transaction the law does not recognize in the form people usually imagine.
Then the investment argument. Permanent plantings take seven to ten years to come into bearing and are financed against decades of expected deliveries at a stated rate. Families committed capital across generations against a published number. Repricing that number is not a market correction to them; it is a retroactive change to the terms of a long contract.
And Proposition 218 was not written to protect irrigation districts. It was written to stop public agencies from levying taxes and calling them fees. The cost-of-service standard that pins IID’ rate at $20 is the same standard that stops a water district anywhere in California from turning a monopoly on a necessity into general revenue. You do not get to keep the protection and discard the constraint.
Last, the scale. A market-clearing price for Imperial Valley water would move a regional agricultural economy to coastal cities in roughly one irrigation season, and the valley’ unemployment rate was 19.7% in July 2026, not seasonally adjusted, before anyone tried it. “Let it clear” is not a small ask, and the people asking are mostly not the ones who would absorb it.
I find all of that serious. What I do not think it survives is the next part.
The arbitrage does not disappear. It changes shape
This year a developer applied to IID for 880 acre-feet a year to cool a data center, and was denied. Not on price — on siting. The parcel sits inside the City of Imperial’ sphere of influence near municipal water infrastructure, so a district service rule pointed the applicant at the city instead.
What the developer did next is the part worth keeping. It bought a 160-acre farm next door, proposed to stop irrigating it, and offered to run the data center on the water the farm had been receiving.
Look at what that choice is actually pricing. The district’ policy for new non-agricultural projects carries a posted, tiered fee schedule; a project at 880 acre-feet a year lands in a tier that works out to roughly $661 an acre-foot in reservation and development fees. The agricultural route to the identical water is $20.
So the developer did nothing exotic. It found the cheaper tier and bought the asset that carries it. Farmland in that valley is the delivery mechanism for a water entitlement priced at a legal floor, and if you want the floor you have to buy the dirt that comes with it. One of the eight claims in the resulting lawsuit is a Proposition 218 challenge to the industrial fees, which is its own kind of confirmation: you can only argue a fee is unlawfully high if a price exists to argue about.
What a capped price does to a hard asset
This is the part I keep turning over. A price ceiling does not destroy the value it caps. It relocates where that value gets expressed.
If an acre-foot is worth $700 to a city and the law says a grower may be charged $20 for it, the missing $680 does not evaporate. It attaches to whatever asset carries the right to the cheap tier — the land, the crop pattern, the position in the district’ queue. The value is still there. It is just no longer in the price, which means it no longer does any of the work a price does: it does not signal, it does not ration, and it does not tell you what the thing is worth without a lawsuit.
I looked at farmland as an inflation hedge in a separate piece on 55 years of USDA land values and found it beat consumer prices by a wide margin while badly lagging growth in the money supply. I was tempted to use this as the explanation and I do not think it holds. The national land-value average is dominated by rain-fed Midwest ground with no district, no tariff, and no acre-foot to speak of. Whatever is behind that lag nationally, it is not Proposition 218.
What the Imperial case does show is narrower and, to me, more useful: an asset can be perfectly scarce, perfectly real, and still fail to carry its own value in its price, if a rule stands between the owner and the highest bidder. Scarcity is necessary. It was never sufficient. That is also the unglamorous reason central banks hold gold rather than something scarcer: what they are buying is the ability to sell it to anyone, on short notice, at a price.
Two things I could not settle, and the test I would watch
I cannot tell whether the development fee in the non-agricultural schedule is one-time where the reservation fee recurs. The names suggest it, the policy page does not say so outright, and if it is true then the $661 figure overstates the annual cost and understates the up-front one. That distinction matters to anyone actually modeling the decision and I would not quote the number without it.
I also do not know what the two existing non-agricultural agreements priced at. Those are the only two real observations of what this water fetches under contract, and they are the comparison I most want and do not have.
The test is already running. That same policy designates up to 25,000 acre-feet a year for new non-agricultural projects. Two agreements and one conditional letter have been issued, for 6,380 acre-feet. That leaves 18,620 uncommitted, and whatever the next agreement prices at is the closest thing to a market signal this system is capable of producing.