Coase Theorem

Two sides can both be describing the same friction honestly. Whether it gets fixed depends on something else entirely: how cheap it is for them to actually strike a deal.

September 14, 2026 8 min read

A support lead says the real problem is sales overpromising features that don't exist yet. The sales lead says the real problem is support refusing to commit to timelines anyone can sell against. Get both of them in a room and each account holds up: sales really does promise things too early, and support really does hedge every estimate into uselessness. The instinct is to figure out who's more wrong and fix that side. That instinct is aimed at the wrong target.

Avoiding one team's harm always means causing the other's

Every friction between two teams looks, from either side, like the other team causing a problem. But the shape of the situation is reciprocal, not one-directional. To stop sales from overpromising, you'd have to slow down or block deals until support signs off, which costs sales real revenue and speed. To stop support from hedging, you'd have to make them commit to estimates before they've actually scoped the work, which costs support real accuracy and credibility. There's no version of "fix the problem" that doesn't impose a cost on someone. The actual question was never who's causing the harm. It's which cost is smaller, and who's positioned to absorb it most cheaply.

That reframing changes what "the fix" even means. It isn't a verdict about which team was behaving badly. It's a trade: does the value support gets from more accurate estimates exceed what sales loses from more upfront review, or does it run the other way? Once the question is a trade instead of a verdict, the obvious next question is whether the two sides can actually strike that trade, and that turns out to depend on something that has nothing to do with who's right.

The theorem that carries a name its author never used

Ronald Coase laid this out formally in "The Problem of Social Cost," published in The Journal of Law and Economics in October 1960. His running example wasn't a support-versus-sales dispute, it was a farmer and a neighboring cattle-raiser. Coase built an arithmetic case: fencing the farmer's property costs $9 a year, each additional steer in the herd causes a specific amount of additional crop damage, and crops sell at $1 a ton. He then worked the numbers twice: once assuming the cattle-raiser is legally liable for crop damage, once assuming he isn't. In both versions, as long as the farmer and the cattle-raiser can actually negotiate with each other at no cost, the herd ends up the same size and the same amount of cultivation happens either way. What changes between the two legal rules isn't the outcome, it's only who ends up paying whom to get there.

That's the result people now call "the Coase theorem": when bargaining between two parties is costless, the efficient outcome doesn't depend on who holds the legal right to begin with, only the distribution of money does. But Coase himself never used that phrase in the paper. The name came six years later, when the economist George Stigler coined "the Coase theorem" in the 1966 edition of his textbook The Theory of Price. Multiple economic historians studying the paper's reception, including Steven Medema's direct examination of how the label took hold, have pointed out that Stigler's name arguably points at the wrong part of Coase's own argument. Coase's actual purpose in the paper wasn't to describe an idealized zero-cost world. He was using that idealized case as a deliberately unrealistic setup, precisely so he could spend the rest of the paper on the world where bargaining is never free, which is the world he actually cared about.

The honest limit: this is a benchmark case, not a description of the world

The zero-transaction-cost result gets invoked, casually, to argue that markets sort out disputes on their own and rules or defaults barely matter. That's a real misreading of what Coase was doing. He stated plainly that the reciprocal-harm cases he analyzed were meant to clarify the logic of the choice, not to claim real bargaining is free. In the real world, finding out what's actually at stake costs something, negotiating a change costs something, and making a new arrangement stick instead of quietly reverting costs something. Coase's own conclusion was the opposite of "rules don't matter": because those costs are always positive, which side holds the default right, and how expensive it is to renegotiate away from that default, ends up mattering enormously. Daniel Kahneman, Jack Knetsch, and Richard Thaler tested this directly in a 1990 experiment, titled in part after the theorem itself. When they ran a market in tokens with an assigned cash value, trading volume matched what the theorem predicted almost exactly, confirming that transaction costs weren't the obstacle in their setup. When they ran the identical market design with real coffee mugs instead of tokens, trading volume came in well below the predicted level every time, driven by the endowment effect rather than by any transaction cost at all. Even in a setting built to approximate the zero-cost case, actual human valuation can still keep the predicted trade from happening. This post isn't a claim about how any particular company should assign ownership of a dispute like the sales-versus-support one above. It's a claim about which variable actually determines whether a fix happens at all.

Where this fits at Brief

Coase's argument only holds together because he named the specific costs that block a bargain once you leave the zero-cost case: the cost of knowing what's actually at stake, the cost of negotiating a change, and the cost of making that change hold instead of drifting back to the old default. Those three costs are exactly what a decision-capture layer is built to cut, not context in some general sense.

Take a pricing default a PM set six months ago, with a real reason behind it, before that PM moved to a different team. Someone on support now wants to change it. Without a record of the original decision, changing it means re-deriving the entire original argument from scratch, tracking down who made the call and why, which is Coase's information cost paid in full every single time the question comes up again. That reinvestigation is also where negotiation cost bites: without a record to point to, proposing the change means reopening the whole original conversation with whoever still remembers it, re-litigating a decision that was already settled once, rather than a quick confirm-or-supersede exchange against a recorded argument. Even if that negotiation happens once and the change gets made, without a durable record marking the old default as superseded and why, the change is fragile: the next person who hits the same edge case has no way to tell a deliberate update from an accidental drift, and the org either re-litigates it again or silently reverts to the old behavior, which is Coase's enforcement cost showing up as regression. A decision captured with its reasoning, and a supersession recorded when it changes, is a direct reduction of all three specific costs Coase identified as the actual obstacle to an efficient renegotiation, not a proxy for "more context is good."

Look at the last cross-team dispute your org argued about instead of fixed. Was the disagreement really about who was right, or about the fact that renegotiating the boundary cost more than either side was willing to spend?

Frequently asked questions

What is the Coase theorem? The result that when two parties can bargain with each other at no cost, the final, efficient allocation of resources doesn't depend on who holds the legal right at the start, only the distribution of money between them does. Change who's liable and the payments change; the underlying outcome doesn't.

Did Coase himself call it "the Coase theorem"? No. His 1960 paper never uses that phrase. The economist George Stigler named it "the Coase theorem" in the 1966 edition of his textbook The Theory of Price, six years after Coase's paper was published.

What was Coase's actual example? A farmer and a neighboring cattle-raiser, worked through with real numbers: $9 a year to fence the property, $1 a ton for crops, and a specific crop-damage cost for each additional steer. Coase showed that whether the cattle-raiser is legally liable for the damage or not, the herd ends up the same size either way, as long as the two of them can negotiate freely. Only who pays whom changes.

Does this mean regulation or clear ownership rules don't matter, since the market sorts it out anyway? No, and this is the most commonly missed part of the paper. Coase's zero-cost case was a deliberate simplification to set up his real argument: in the real world bargaining is never free, so where a default right sits, and how cheap it is to renegotiate away from it, determines the outcome far more than any assumption that people will just work it out.