Over the last two decades, the fashionable instrument of environmental policy has been the market mechanism. Cap and trade, output based pricing, low carbon fuel standards, offset markets: each was sold on the same premise, that a price left to find its own level would allocate abatement more efficiently than a bureaucrat ever could. Let the market discover where the cheapest tonne lives, and it will be found and cut. It is a genuinely good idea, and it rests on a genuinely deep one: that a price aggregates dispersed information no central planner can assemble.

The trouble is what has happened since. Program by program, amendment by amendment, governments have been fitting these markets with controls. A ceiling here to stop prices spiking, a floor there to stop them collapsing, a quantity adjustment to rebalance supply, an eligibility rule to steer which credits count, a multiplier to protect a favored producer. Each control is added for a defensible reason. Cumulatively, they are converting instruments that were meant to discover prices into instruments that administer them, with a trading layer left on top to preserve the appearance of a market. The result is a slow drift from market toward mandate, and it is worth naming because the name on the tin still says market.

What a market is supposed to do

The case for a market instrument is not that trading is intrinsically virtuous. It is about knowledge. A price that moves freely carries information: it tells thousands of dispersed actors, none of whom can see the whole system, where value and scarcity actually sit, and it lets each of them respond to knowledge they alone hold. That is the entire point. Strip out the discovery and you have not kept the good part of a market and shed the messy part. You have kept the messy part, the transaction costs and the chain of steps in conducting a transaction, and shed the good part, the information.

A command and control system does the opposite. A regulator sets the quantity, or the price, or the technology by administrative decision, and in doing so discards the very thing that made the market worth using: the information. It substitutes one office's estimate for the dispersed judgment of everyone who would otherwise have traded. For a narrow problem with a single obvious answer, banning a specific toxin, that trade can be worth making. But for a problem as vast and dispersed as decarbonizing an entire economy, where the cheapest tonnes are hidden inside tens of thousands of decisions no ministry can see, central planning is close to the worst tool available, precisely because it throws away the knowledge the task most requires. A planner setting prices by decree will set them wrong, and the whole economy pays the gap between the guess and the truth.

So the ranking here is not a matter of taste, and it is worth stating plainly. For a problem like this one, a genuine market is the best instrument we have and command and control is close to the worst, and the distance between them is exactly the information a free price carries and an administered one cannot. That is what makes the drift worth worrying about. The danger is not that governments choose planning openly and defend it. It is that they choose a market, and then manipulate it so continuously that they arrive at planning anyway, having thrown away the market's one irreplaceable advantage while keeping all of its cost and machinery. That is not a middle path between the two. It is the worst tool wearing the costume of the best, and the costume is what stops anyone from noticing the trade was made.

The accumulation of governors

Think of each control as a governor, the mechanical device that caps how fast an engine can run. Look at the modern carbon market and count them.

Most cap and trade systems now carry an auction reserve price, a floor below which allowances will not sell, and a cost containment reserve or hard ceiling that releases supply or caps cost when prices climb. Between the two, the price is fenced into a lane. Regulators reserve the right to adjust the number of allowances when the balance drifts, which is a direct hand on the supply curve. Low carbon fuel standards in California and Washington spent the last two years being rewritten specifically because credits had piled up and prices had fallen, Washington credits sliding from roughly 107 dollars at launch toward the mid twenties, and the fix was to tighten the benchmark, which is to say the regulator adjusted quantity to move price. Offset markets run eligibility lists that determine, by decision, which tonnes are allowed to compete.

Every one of these is reasonable in isolation. A price spike is politically intolerable and can bankrupt compliance entities. A price collapse kills the environmental signal and the investment thesis with it. Oversupply makes the program look like a giveaway. So the regulator reaches for a governor, and the governor works. But stack enough of them and the price is no longer discovering anything. It is expressing the space the regulator has left it, bounded above by a ceiling the regulator set, below by a floor the regulator set, and pushed around in between by quantity the regulator controls. The trades are real and the money is real. The discovery is mostly gone.

The Clean Fuel Regulations, as a current example

Canada's Clean Fuel Regulations are a vivid instance, because the government has effectively said the quiet part in writing.

Facing a competitiveness shock after the United States restructured its biofuel subsidies, Environment and Climate Change Canada proposed in late 2025 to amend the CFR to protect domestic producers. One option is a credit multiplier: award a Canadian litre of low carbon fuel more credits than an identical imported litre. The paper does not calibrate the multiplier to carbon intensity or to the cost of abatement. It back solves it from the size of the American subsidy, working an example where a multiplier of 1.4 would match the US per litre incentive for renewable diesel and 1.14 for ethanol. The number of credits a fuel earns is chosen to hit an income target for a domestic industry.

Then comes the admission. ECCC notes that these extra credits would not represent incremental emission reductions, and could even reduce demand for the physical low carbon fuel in the system, since suppliers could meet their obligations with fewer litres. A credit that corresponds to no additional tonne is not a commodity trading on discovered value. It is a subsidy voucher denominated in credits. And the discrimination by origin, treating an otherwise identical litre differently based on where it was made, is the signature of trade policy, not carbon policy, because a tonne abated is worth the same wherever it happens. The CFR here is not drifting toward command and control by accident. It is being consciously steered there, with the credit repurposed as a delivery mechanism for industrial support.

Output based pricing is the cleaner case

The CFR is dramatic, but output based systems like the federal Output Based Pricing System and Alberta's TIER show the drift in its purest form, and they matter more because they cover heavy industry.

In these systems the central lever is the benchmark, the emissions intensity standard a facility is measured against. Beat it and you earn credits, miss it and you owe. The benchmark is not a market outcome. It is set, and periodically recalibrated, by the regulator. And because the benchmark determines how many facilities are long and how many are short, it effectively sets the scarcity of the whole credit pool before a single trade occurs. Tighten the benchmark and you manufacture demand for credits. Loosen it and you flood the market. The price that emerges is a genuine clearing price, but it is clearing against a scarcity the regulator dialed in. This is the thing worth sitting with: in an output based market, the most important price determining decision is an administrative one made upstream of the market, and the market's job is reduced to distributing the consequences. The trading is real. The scarcity it prices is designed.

That is not a criticism of TIER, which functions and delivers reductions. It is an observation about what kind of object it is. When the regulator sets the benchmark, holds a price ceiling, and can adjust the rules, the instrument is closer to a regulated utility with a tradable overlay than to a commodity exchange. We keep calling it a market because it has prices and trades. Both can be true of an administered system.

Why the drift is not an accident

It would be easy to read this as regulators failing to keep their hands off. It is closer to the opposite. The drift is structural, and it follows from a simple political fact: a genuine market outcome includes prices that elected governments cannot tolerate.

An unfenced carbon price spikes when abatement is scarce and collapses when it is abundant. The spike lands on consumers and trade exposed industry as a visible cost, and it arrives on a political timetable no one controls. The collapse guts the environmental goal and strands the investments made in good faith. Both are, in the language of markets, the price doing its job, transmitting real information about scarcity. Both are, in the language of politics, a crisis. No government can credibly promise to leave the price alone through either, and everyone in the market knows it, which means the implicit governors are priced in even before they are used. The market is administered in expectation, not only in fact.

So the accumulation of controls is not a bug that better discipline would fix. It is what happens when you ask a market to produce an outcome a political system needs to keep inside a range. The mechanism and the mandate are pulling in different directions, and over time the mandate tends to win, one reasonable amendment at a time. That is exactly why the drift has to be named and resisted rather than waved through as a series of technical fixes. The pressure only ever runs one way, toward more control, and a market is not destroyed in a single stroke. It is dismantled by a thousand defensible cuts, each of which is easy to justify and none of which is ever framed as the choice it actually is: another step back toward the worst tool for the job.

The honest counterargument

The strongest reply to all of this is that the governors are worth their cost, and that a bounded market still does real work. That is fair, and it is true in specific places. Even inside a fenced program, the market genuinely discovers some things. It prices the relative value of a low carbon intensity credit against a high one. It finds the cheapest compliance path within the lane it is given. And it discovers prices most freely not inside any single program but between them, in the arbitrage that flows when a participant can satisfy one obligation with credits stacked from another. The discovery did not vanish. It migrated, to the margins and to the seams.

One can also argue that a purely free carbon price was never on offer, that the realistic choice was always between an administered instrument that trades and an administered instrument that does not, and that the trading version is more efficient even heavily governed. That may well be right, and it is not a reason for despair so much as a reason for precision. Even if a perfectly free price was never realistic, far more discovery was on offer than these programs now permit, and the live choice at every margin is still whether to protect what remains of it or to legislate it away. The gap between how these instruments are marketed and how they are run is exactly what lets that choice get made, again and again, without anyone naming it as a choice.

The question worth asking

This is not an argument that a market should run entirely without guardrails. A narrow control that keeps the mechanism alive through a genuine shock, a circuit breaker rather than a steering wheel, can be exactly what preserves a functioning market when it would otherwise fail. But there is a world of difference between a guardrail that keeps a market working and the continuous manipulation that quietly replaces it, and the two are constantly conflated. Every intervention is defended as the first. In aggregate they deliver the second. The point is not that controls are always wrong. It is that the goal has to be a working market, and most of what is being added no longer serves that goal. It serves the outcome the regulator wanted before the market opened, which is planning by another name.

The honest test is simple. Does the price still discover something the regulator did not already decide? Where the answer has quietly become no, the instrument is no longer a market in any sense that matters, whatever the name on the tin, and continuing to call it one is not harmless. The label is what lets the drift proceed unchallenged, because a market that is failing invites scrutiny while a market that has merely been converted into a plan and left its sign up does not.

It also points somewhere constructive. If the internal price of any single program is increasingly a policy output rather than a discovered signal, then the honest place to locate genuine market discovery is between programs, at the layer where credits of different origins, qualities, and risk profiles are made comparable and exchanged. The scarcity inside each silo is designed. The relative value between silos is not, or need not be. Whatever the future of any individual program, that is the layer where a market can still do the one thing only a market can do, which is discover a price that no one set.