The EU’s carbon market was designed to eliminate emissions. It built a financial industry that needs them to continue.
How it’s supposed to work
Governments have two basic ways to reduce industrial pollution. They can regulate directly: set emission limits for each factory, mandate specific clean technologies, ban the dirtiest processes. Or they can create a market. That’s what the EU did in 2005 with the Emissions Trading System.
The idea is this. The government sets a ceiling on total greenhouse gas emissions. Every tonne of pollution requires a permit. Companies that want to pollute must hold enough permits to cover their emissions. The ceiling drops every year. Permits get scarcer. Their price rises. Companies that clean up can sell their spare permits to companies that haven’t, so the reductions happen wherever they’re cheapest. By 2039, the ceiling hits zero. No more permits. No more emissions. The system eliminates itself, and the problem along with it.
The reason to use a market instead of regulation is cost. A government regulator can’t know which of the thousands of factories across Europe can cut emissions most cheaply. But if you let companies trade permits, the market figures it out: the company that can cut cheaply does so and sells its permits at a profit; the company that can’t, buys permits instead of making expensive changes it can’t afford. Total emissions stay the same. Total cost goes down. That’s the theory.
For twenty years, this has been Europe’s flagship climate policy. It covers power stations, steel mills, cement plants, airlines, and shipping, roughly 40% of EU emissions. It has generated over €245 billion in government revenue from permit auctions.
It is also an €881 billion annual trading market where, as of 2024, investment firms and banks account for 63.5% of all volume, 72% of trades come from outside Europe, each permit changes hands ten times on average before it’s actually used to cover emissions, and 909 institutions hold derivatives positions on any given day.
The system is still an emissions mechanism on paper. But in practice, it has developed the structure, the participants, and the incentive dynamics of a financial market. And financial markets do not build themselves toward their own extinction.
How did an environmental policy become a financial ecosystem? Through four design choices, each economically logical, each politically convenient, and each producing consequences that the designers never modelled.
Four choices that broke in practice
“It doesn’t matter whether you auction the permits or give them away free. The market reaches the same outcome either way.”
This is a real theorem. In 1960, Ronald Coase showed that when property rights are clearly defined and transaction costs are zero, parties negotiate to an efficient outcome regardless of who holds the initial rights. In 1972, W. David Montgomery formalised this for pollution markets, proving that transferable permits achieve cost-effective control. By the time the ETS was designed, it had become conventional wisdom: allocation method doesn’t affect emissions; the cap determines the outcome, the market just finds the cheapest path. The maths checks out, under idealised conditions.
It gave politicians a gift. They could tell industry: your permits are free, no cost to you. And tell environmentalists: the emissions outcome is identical. Both statements were technically true, under the model’s assumptions.
What the model didn’t account for was what happens when you hand companies something worth billions — for free, year after year. Between 2013 and 2021, European heavy industry received €98.5 billion in free permits. That’s more than the €88.5 billion governments collected from auctioning permits in the same period. Industry got more from the system than the system raised.
Companies didn’t just use these permits. They accumulated surpluses and sold them. ArcelorMittal made €1.9 billion from surplus sales alone between 2005 and 2019. LafargeHolcim earned nearly €1 billion. On top of that, companies passed through the theoretical cost of carbon to their customers, charging as if they’d paid for permits they’d received for free. Total windfall profits across 15 sectors in 19 countries: €30 to over €50 billion.
In 2023 alone, ArcelorMittal received more than €3.8 billion worth of free permits. Heidelberg Materials received nearly €2 billion. And in 2024, 15 national governments paid out €5.52 billion in additional subsidies to cover indirect carbon costs for energy-intensive industries, a 40% increase over the previous year.
The theorem said allocation method doesn’t affect the outcome. It didn’t model that free allocation creates a constituency: companies with billions of reasons to lobby for the system to continue exactly as it is. The subsidy built the lobby that defends the subsidy. The theorem was correct. The assumption that politics wouldn’t interfere with the theorem was not.
“The cap guarantees the environmental outcome.”
Set the ceiling, let the market find the cheapest path beneath it. As long as emissions can’t exceed the cap, the target is met by design.
Three things broke this.
First, reliable emissions data didn’t exist when the system launched. Phase 1 caps were set through National Allocation Plans, where each member state estimated its own ceiling under heavy industry lobbying for generous limits. The European Commission’s own review is blunt: “In the absence of reliable emissions data, phase 1 caps were set on the basis of estimates.” The result: companies received more free permits than they actually emitted — before reducing anything. The price collapsed to near zero because there was nothing to comply with.
Second, the cap only covers 40% of EU emissions. Inside the cap, emissions fell 50%. Outside it (agriculture, buildings, most transport), only 20%. The system reduced emissions sharply in the sectors it covered and barely touched the 60% it didn’t.
The exclusions aren’t random. Agriculture is the EU’s single largest source of methane, 80 times more potent than CO₂ over twenty years. It’s excluded from the ETS. It’s excluded from ETS2. The EU Methane Regulation, adopted in 2024, covers only the energy sector. The official reason is measurability: farm emissions are diffuse. But satellite-based methane detection can now identify emissions at facility level. What remains is the political reality: the Common Agricultural Policy is €387 billion. The tractor protests of 2024 forced the Commission to retreat on pesticide and nature restoration rules. No government wants to be the one that made milk more expensive because of an emissions permit.
And even within the boundary the cap claims to cover, a significant share of emissions go uncounted. The ETS treats biomass burning as zero-emission at the smokestack, on the assumption that the trees absorbed the same CO₂ while growing. The European Commission’s own 2025 Carbon Market Report shows that these zero-rated biomass emissions amount to 22% on top of what ETS installations actually report. But the carbon debt of the forests being logged to feed European power plants can take centuries to repay. The emissions are real. The accounting says they’re zero.
Third, and most importantly: the cap reaching zero depends on it being politically irrevocable. It isn’t. Germany’s environment minister has publicly urged extension beyond 2039. The European Commission is reportedly planning continued industrial emissions after that date. The cap was designed as a technical parameter. It functions as a political variable, one that every actor in the system has an incentive to adjust.
“The carbon price drives investment in clean alternatives.”
The logic: companies see the price of pollution, do the maths, invest in cleaner technology when it’s cheaper than buying permits. Higher price, more investment. The market discovers the optimal path.
The problem is that 90% of industrial emissions are covered by free allocation. In theory, free permits still carry an opportunity cost — every permit you use is one you could sell at market price. Rational firms should treat them identically to purchased permits. And some do: the windfall profits described above come precisely from companies passing through this opportunity cost to customers.
But that pass-through shows why the theory breaks down in practice. A cement company that receives free permits, charges customers the carbon cost, sells the surplus, and books the profit never writes a cheque for its pollution. The carbon price shows up on its balance sheet as revenue, not as a cost. Economically, the incentive to reduce should be the same. Behaviourally, it isn’t — because eliminating your emissions also eliminates the free permits, the surplus sales, and the pass-through revenue that come with them.
The evidence is in the split. In 2013, the power sector lost its free permits and had to buy at auction. Since then, electricity emissions dropped 28.6%. Industrial emissions, still shielded by free allocation, dropped less than 9%. Where companies pay for their pollution, they reduce it. Where they don’t, they don’t.
Picture a cement plant operator in Germany. She receives her emissions permits for free, benchmarked against sectoral averages. She passes the theoretical carbon cost through to her customers — revenue for a cost she never paid. She has surplus permits she can sell on the market. Now someone proposes she invest €500 million in a breakthrough low-carbon kiln. The investment would eliminate her emissions — and with them, her free allocation, her surplus sales, and the pass-through revenue. The system is asking her to spend half a billion euros to destroy her own subsidy.
There’s a deeper issue, too. Building that kiln is a 15-20 year capital decision. Carbon prices have swung from under €5 to over €100 in the system’s lifetime. That volatility is profitable for traders. It’s devastating for anyone trying to build a business case for a billion-euro industrial transformation. What decarbonisation investment needs is a credible, stable, rising price floor. What the ETS provides is a volatile market price, subject to political intervention whenever it gets high enough to bite.
“The market is just a tool. When the job is done, the tool is no longer needed.”
This assumption was never stated because it seemed too obvious to need stating. The ETS is an environmental policy that uses a market mechanism. The market is a means, not an end. When emissions reach zero, the mechanism becomes unnecessary.
Then in 2018, EU emission allowances were classified as financial instruments under MiFID II, for the first time bringing them under the same regulatory regime as stocks and bonds. At that moment, the tool formally became a financial market. It attracted the participants that financial markets attract: investment firms, credit institutions, hundreds of investment funds, hedge funds. It developed the infrastructure that financial markets develop: futures, options, clearing houses, trading desks, data services, compliance platforms, advisory practices.
€881 billion in annual trading volume. Each permit transacted ten times. 909 daily derivatives position holders. Trading platforms telling their investors that the carbon market is here to stay.
The tool became an institution. And institutions do not design themselves toward obsolescence.
The dynamic
None of this required conspiracy. It required only incentive alignment.
Industry receives free permits worth billions. It lobbies to keep them, first supporting the Carbon Border Adjustment Mechanism as a replacement for free allocation, then reversing position as the phase-out date approached. Industry groups now push to delay the transition until CBAM is “proved fully effective” — a condition no one has defined and that could justify indefinite postponement.
The financial infrastructure needs the market to persist. €881 billion in annual trading doesn’t unwind quietly. The traders, the clearing houses, the exchanges, the compliance consultants: none of them need to manipulate anything. They just need to exist at sufficient scale that their disruption becomes politically unthinkable.
When carbon prices spike, the political response isn’t to let the price signal work. In 2021, when prices surged past €90, Poland and Spain demanded restrictions on financial participants. The Commission asked ESMA to investigate market manipulation. ESMA found nothing abnormal. But the political pressure didn’t disappear — it moved. When ETS2 was set to launch in 2027, covering buildings and transport for the first time, governments delayed it to 2028 over concerns about consumer costs. Every time the system threatens to impose real costs, the response softens it.
And through it all, the headline number provides cover. Emissions in covered sectors are down 50% since 2005. The 2030 target of 62% reduction appears achievable. Economists cite studies showing the ETS caused emissions reductions of between 8% and 16%, depending on the study and methodology. These findings are correct but misleading. The reduction is overwhelmingly driven by the power sector, where free allocation was removed. Industrial decarbonisation, the expensive part, the hard part, has barely moved. But the aggregate number supplies legitimacy for a system whose defining feature is its exemption regime.
Nobody was corrupt. The carbon leakage argument (that industry would relocate to unregulated countries) had economic logic. The benchmarking system attempted to reward the most efficient producers. Politicians faced genuine competitiveness concerns. Every decision was individually defensible. The structural outcome was that Europe’s heaviest polluters were paid billions to continue polluting while the system’s headline numbers claimed progress.
The endgame paradox
Under current legislation, the cap reaches zero by approximately 2039. No new permits. When existing surplus is exhausted, no further emissions permitted. This is the system’s stated destination. It is also the trading industry’s extinction event.
A system with zero cap has zero supply. No permits to auction, no futures to write, no positions to take. The financial infrastructure that accounts for 63.5% of trading volume has no product. The compliance industry has no compliance to manage. The platforms have no trades to intermediate.
And here is where the system’s self-preservation becomes visible. The calls to extend beyond 2039 aren’t coming from the margins — they’re coming from Germany’s environment minister and the European Commission itself. The 2026 ETS revision, the most consequential reform since the system’s creation, will determine whether the cap trajectory holds.
The lobbying frame is competitiveness. The structural reality is that every actor in the system has a rational interest in decarbonisation proceeding at precisely the pace that sustains the system. Industry receiving free permits. Financial firms trading them. Consultants advising on compliance. Regulators administering the mechanism. Politicians spending auction revenues. All aligned. Fast enough to claim it works. Slow enough that it never reaches zero.
Optimisation vs. transformation
Buying permits, selling the surplus, using futures contracts to manage the price risk — that’s optimisation. You’re navigating the system as it exists. You shave a percentage off emissions, profit from the rest.
Building a green steel plant that eliminates the need for permits entirely — that’s transformation. Stegra in Sweden (formerly H2 Green Steel) isn’t hedging its carbon price. It’s rendering the carbon price irrelevant by removing the emission at source.
The distinction matters because the ETS, by design, rewards optimisation and punishes transformation. A company that eliminates its emissions loses its free allocation, exits the trading ecosystem, and gives up the revenue from surplus permit sales. A company that reduces just enough to generate surplus gets to sell those permits for profit while maintaining its position in a system that subsidises its participation.
Denmark’s cooperative energy model shows what the alternative looks like. By the late 1990s, over 100,000 families owned shares in wind turbine cooperatives that had installed 86% of the country’s wind capacity. They didn’t trade a carbon permit. They built the infrastructure and became a political constituency for keeping it. The Mankala model in Finland, where industrial consumers co-own power generation at cost, doesn’t need a carbon market to incentivise clean energy. The ownership structure makes the market unnecessary.
These models don’t reduce emissions by trading permits more efficiently. They eliminate the need for permits by building the clean infrastructure directly.
The EU ETS is the most sophisticated emissions optimisation system ever built. It has generated over €245 billion in revenue. It employs thousands. It supports academic research, policy careers, trading desks, and consulting practices across the continent.
It has also, after twenty years and nearly a trillion euros in annual trading volume, delivered less than 9% industrial emissions reduction while handing the same industries €98.5 billion in free pollution permits and generating windfall profits of up to €50 billion.
Every design assumption was economically correct in theory. Every one broke in practice — not because the economics were wrong, but because the economics didn’t model the political economy that grew around the system itself. The theorem held. The world around the theorem did not.
At some point, the question isn’t whether the system is working. It’s what the system is working for.
Meanwhile
While Europe debates extending its carbon market past 2039, the physical world is not waiting for the timeline.
The 2025 Planetary Health Check found that seven of the nine planetary boundaries that define a safe operating space for human civilisation have now been breached. Climate change, biodiversity loss, deforestation, freshwater use, agricultural pollution, synthetic chemical contamination, and — for the first time in 2025 — ocean acidification. All seven are worsening. Only ozone depletion and aerosol loading remain within safe limits.
Atmospheric CO₂ hit 430 parts per million in May 2025, the highest level in at least 800,000 years. The planetary boundary for climate is 350 ppm. We passed it decades ago and are accelerating away from it. The 2025 Global Carbon Budget estimates that the remaining carbon budget to limit warming to 1.5°C will be exhausted before the end of the decade at current emissions. Fossil fuel emissions set another record high in 2025.
The ETS was designed in a world that assumed it had time. Time to let the cap decline gradually. Time to let the market discover the efficient path. Time to let the political system absorb the costs incrementally.
The remaining carbon budget for 1.5°C runs out before 2030. The ETS cap reaches zero in 2039.
That gap is not a rounding error. It is the distance between optimisation and survival.
References
Market structure and trading data
ESMA, Annual Report on EU Carbon Markets (2024 edition, October 2024; 2025 edition, October 2025). Source for: 63.5% trading volume by investment firms/credit institutions, 72% on-exchange volume from non-EEA entities (36% US, 25% UK), 909 daily derivatives position holders (2024 data), 453 investment funds holding 6% of positions, futures share of trading. ESMA 2022 preliminary and final reports on speculation found no abnormality in market functioning; speculative positions ~4% of market.
Verifavia / Refinitiv, Annual EU ETS Carbon Market Report 2024 (2025). Source for: €881 billion total trading volume (2023), 9.7 billion EUAs traded (2024), each EUA transacted ~10 times, 85% futures contracts.
European Commission, EU ETS Factsheet — Auctioning (2025). Source for: over €245 billion in auction revenues since 2013, €39 billion in 2024 revenue, 62% reduction target by 2030.
Free allowances, windfall profits and industrial emissions
Carbon Market Watch, Free Pollution Permits: Exposed (2025). Source for: €98.5 billion free allowances to industry 2013-2021, €88.5 billion auction revenue same period, ArcelorMittal €1.9 billion surplus sales 2005-2019, LafargeHolcim nearly €1 billion, ArcelorMittal €3.8 billion free allowances 2023, Heidelberg Materials nearly €2 billion 2023, 90% industrial emissions covered by free allocation, 28.6% power sector reduction vs. less than 9% industrial reduction since 2013, €5.52 billion indirect cost compensation 2024 (40% increase), CBAM lobbying reversal.
CE Delft (de Bruyn et al.), Additional Profits of Sectors and Firms from the EU ETS 2008-2019 (2021), commissioned by Carbon Market Watch. Source for: total windfall profits €30 to over €50 billion across 15 sectors in 19 countries.
Mramor & Tagliapietra, Europe’s Emissions Trading System is an Ally, Not an Enemy, of Industrial Competitiveness, Bruegel Analysis (January 2026). Corroborates windfall profits range (€26-€46 billion for cost pass-through component) and industrial emissions reduction of less than 9%.
Emissions reductions and ETS performance
European Commission, 2025 Carbon Market Report (2025). Source for: 50% emissions reduction since 2005 in covered sectors, 62% 2030 target on track, 20% reduction in non-covered sectors.
Bayer, P. & Aklin, M., The European Union Emissions Trading System Reduced CO₂ Emissions Despite Low Prices, Proceedings of the National Academy of Sciences, 117(16), 8804-8812 (2020). Source for: ETS reduced emissions by 8.1-11.5% in covered sectors (2008-2016).
Colmer, J., Martin, R., Muûls, M. & Wagner, U., Does Pricing Carbon Mitigate Climate Change? Firm-Level Evidence from the European Union Emissions Trading Scheme, Review of Economic Studies (2024). Source for: EU ETS induced 14-16% emissions reduction in regulated manufacturing firms.
European Environment Agency, Greenhouse Gas Emissions Under the EU Emissions Trading System (2025). Source for: ETS covers ~37-40% of EU greenhouse gas emissions, 51% reduction in stationary installation emissions 2005-2024.
European Commission, Development of EU ETS (2005-2020) (2024). Source for: “In the absence of reliable emissions data, phase 1 caps were set on the basis of estimates”; National Allocation Plans process; price collapse to zero in 2007.
Cap trajectory, 2039 endgame and policy extensions
Clean Energy Wire, German Environment Minister Urges EU to Extend Industry Emissions Trading Beyond 2039 (2025). Source for: Carsten Schneider’s call to extend ETS beyond 2039.
Clean Energy Wire / Handelsblatt, European Commission Plans to Propose Extending Free CO₂ Allowances (2025). Source for: Commission reportedly planning to extend free allocation beyond 2034, stretch auctioning beyond 2039.
Enerdata, Carbon Price Forecast Under the EU ETS (2025). Source for: ETS endgame analysis, cap reaching zero timeline, MSR behaviour, price trajectory modelling.
European Environment Agency, Greenhouse Gas Emissions Under the EU Emissions Trading System (2025) and Clean Energy Wire, Understanding the European Union’s Emissions Trading Systems (2025). Source for: ETS2 delayed from 2027 to 2028, price history (under €5 to over €100), cap reaches zero ~2039.
Theoretical framework
Coase, R., The Problem of Social Cost, Journal of Law and Economics (1960). Foundational theory for cap-and-trade allocation independence.
Montgomery, D.W., Markets in Licenses and Efficient Pollution Control Programs, Journal of Economic Theory (1972). Formalisation of cap-and-trade cost-effectiveness.
Zaklan, A., Coase and Cap-and-Trade: Evidence on the Independence Property from the European Carbon Market, American Economic Journal: Economic Policy (2023). Source for: independence property holds for large emitters, fails for small emitters.
MiFID II classification
European Commission, Carbon Market — Questions and Answers (2022); European Commission, Ensuring the Integrity of the European Carbon Market (2025). Source for: EUAs classified as financial instruments under MiFID II from January 2018.
Transformation examples
UNFCCC, Wind Energy in Denmark — Case Study: Good Practices and Lessons Learned (2023). Source for: over 100,000 families in 2,100+ cooperatives owning ~90% of Danish wind turbines by late 1990s.
Gorroño-Albizu, L., Sperling, K. & Djørup, S., The Past, Present and Uncertain Future of Community Energy in Denmark, Energy Research & Social Science, 57, 101231 (2019). Source for: 100,000+ families in cooperatives owning majority of Danish wind capacity by early 2000s.
Agricultural emissions and methane exclusion
European Commission, EU Methane Action Plan (2020). Source for: 206 Mt CO₂-equivalent agricultural methane annually, methane GWP-20 of ~80 per IPCC AR6.
European Commission, Regulation (EU) 2024/1106 — EU Methane Regulation (2024). Source for: methane regulation covers energy sector only.
European Commission / European Parliament, CAP Funds (2021-2027). Source for: €386.6-387 billion total CAP allocation 2021-2027.
Biomass zero-rating
European Commission, 2025 Carbon Market Report, COM(2025) 735 (2025). Source for: zero-rated biomass emissions account for 22% on top of ETS installations’ reported emissions in 2024.
Planetary boundaries and carbon budget
Sakschewski, B., Caesar, L. et al., Planetary Health Check 2025, Potsdam Institute for Climate Impact Research (September 2025). Source for: 7 of 9 planetary boundaries breached.
Scripps Institution of Oceanography / NOAA Global Monitoring Laboratory, Annual Carbon Dioxide Peak (June 2025). Source for: May 2025 monthly average CO₂ of 430.2-430.5 ppm.
Friedlingstein, P. et al., Global Carbon Budget 2025, Earth System Science Data (November 2025). Source for: remaining carbon budget for 1.5°C approximately 170 Gt CO₂ (~4 years at current emissions), fossil fuel CO₂ emissions at record high in 2025.