How an over-supplied pilot in 2005 turned into a credible carbon price, traced through the system’s four trading phases.
The EU ETS began in 2005 as a deliberate “learn by doing” pilot. There was no central cap. Each member state wrote its own National Allocation Plan and handed out almost all of the allowances for free. The trouble was that governments built those plans on cautious guesses about how much industry actually emitted, and most countries had no proper measurement system in place yet. The result was far too many allowances. When the first verified emissions figures for 2005 came out in 2006, they confirmed it: the market was holding more allowances than there were emissions to cover.
The price reaction was swift. EUAs had climbed to around €30 in early 2006, but once the surplus became obvious they slid hard, and because Phase 1 allowances could not be carried over into Phase 2 they were worth almost nothing by the end of 2007. An expensive way to learn a lesson, though the plumbing of the system, the registries, the monitoring and the verification, had at least been built and tested for the phases to come.
Takeaway
Hand out allowances for free, let each country set its own total, and err on the generous side, and you end up with a price worth nothing. Much of what came later was, in one way or another, an effort not to repeat Phase 1.
Phase 2 lined up with the first commitment period of the Kyoto Protocol, and it tightened things up. Caps were lower, allowances could now be banked into Phase 3, and a small slice was sold at auction rather than given away. Prices recovered to the €20 to €30 range in early 2008.
Then the financial crisis hit. Factories went quiet, emissions dropped well below the cap, and a fresh surplus of unused allowances piled up, made worse by a flood of cheap international offset credits that companies were allowed to use. The price drifted down for years, and by 2013 it had fallen below €5. At that level it did little to encourage anyone to invest in cleaner technology, and people began to ask whether the scheme was working at all.
Phase 3 was where the system was rebuilt. The patchwork of national plans was scrapped for a single EU-wide cap, set at 2,084 million allowances in 2013 and falling by 1.74% a year under the new Linear Reduction Factor. Auctioning became the main way allowances were handed out, and the free allowances that remained were tied to EU-wide efficiency benchmarks, so that only the better-performing plants were fully covered. Aviation had been folded in from 2012, just before the phase began.
Even so, the crisis surplus kept the price down, and it took two moves to deal with it. The first was “backloading”, a stopgap that pushed the auctioning of 900 million allowances out of 2014 to 2016 and into later years. The second mattered far more: the EU set up the Market Stability Reserve, a permanent mechanism that soaks up surplus allowances automatically according to fixed rules. In the end the backloaded allowances were placed into the reserve rather than sold back to the market.
The turning point
Once the MSR started operating in January 2019, the market finally had a credible way to drain the surplus. Prices climbed sharply over the same stretch, from single digits in 2017 to the mid-€20s by 2019. For the first time the carbon price was high enough to tip power generators toward gas instead of coal.
Phase 4 opened in 2021 with a steeper LRF of 2.2% and a beefed-up MSR. The bigger change, though, came with the 2023 “Fit for 55” reform, which brought the ETS into line with the EU’s goal of cutting net emissions 55% by 2030. The cap was made to fall much faster: the LRF rose to 4.3% for 2024 to 2027 and 4.4% from 2028, with one-off cuts of 90 million allowances in 2024 and 27 million in 2026 layered on top. All of that lifts the sector’s 2030 target to a 62% cut below 2005 levels.
The same reform widened the system’s reach. Shipping came in from 2024, free allocation began to be phased out as the Carbon Border Adjustment Mechanism took its place, and the groundwork was laid for a second, separate market, ETS2, covering buildings and road transport. With supply tightening, EUAs reached a record near €100 in February 2023 before settling into a €60 to €95 band through 2025 and 2026.
| Phase | Defining Feature | Typical Price |
|---|---|---|
| 1 (2005–07) | National caps, free allocation, no banking | €30 → €0 |
| 2 (2008–12) | Kyoto period; crisis surplus builds | €20 → <€5 |
| 3 (2013–20) | Single EU cap, auctioning, backloading, MSR | €5 → €25 |
| 4 (2021–30) | Fit for 55; LRF 4.3–4.4%; shipping added | €50 → ~€100 → ~€74 |
Prices are indicative ranges for illustration, not precise daily quotes.
The MSR has to react to something, and that something is the TNAC, the Total Number of Allowances in Circulation. The Commission publishes it every May. It is the running tally of the market’s surplus: every allowance ever issued or released, minus the emissions already surrendered against and minus whatever is sitting in the reserve. Put simply, it counts how many spare allowances are floating around that companies have banked rather than handed back.
That single number decides each year’s MSR move, judged against two thresholds. If the TNAC is above 1,096 million, the reserve takes 24% of it out of the auctions over the next twelve months. If it falls into the band between 833 million and 1,096 million, the reserve instead removes only the amount above 833 million, a gentler intake added in 2024 so the supply doesn’t fall off a cliff the moment the surplus dips under the top threshold. And if the TNAC ever drops below 833 million, the flow reverses and 100 million allowances are handed back to the market.
For most of the past decade that number was enormous. The overhang left by the financial crisis and a wave of cheap offsets pushed the surplus to roughly 2 billion allowances around 2013, more than a whole year’s emissions sitting idle. The MSR has been grinding it down ever since. It stood near 1.39 billion for 2019, edged back up to about 1.45 billion in 2021 as the pandemic cut emissions, then fell to around 1.13 billion in 2022, 1.11 billion in 2023 and 1.15 billion in 2024.
| Surplus year | TNAC (approx.) | What the MSR did |
|---|---|---|
| 2013 (peak) | ~2.0 billion | Surplus peaks; MSR not yet running |
| 2019 | ~1.39 billion | 24% withdrawn |
| 2021 | ~1.45 billion | 24% withdrawn |
| 2022 | ~1.13 billion | 24% withdrawn |
| 2023 | ~1.11 billion | 24% withdrawn |
| 2024 | ~1.15 billion | 276 million withdrawn (Sep 2025–Aug 2026) |
| 2025 | ~1.02 billion | 190 million withdrawn (Sep 2026–Aug 2027) |
The 2025 figure, published on 29 May 2026, marked a turn. The 190 million the reserve will pull from September 2026 is exactly the TNAC minus 833 million, which means the surplus has dropped to roughly 1.02 billion, below the 1,096 million trigger for the first time and into the gentler band. After a decade spent soaking up an enormous overhang, the MSR is now working on a surplus that is finally getting close to normal.
Why TNAC matters
TNAC is the dial the whole supply mechanism turns on. A big surplus gives buyers a cushion and keeps the price soft; as it shrinks, that cushion thins and the price starts to track each year’s cap cut far more closely. The May TNAC release is one of the simplest things to watch if you want to read where the market is heading.
Step back and the story is fairly consistent: a market that kept being over-supplied, then tightened, then tightened again. The early phases showed that how you hand out allowances matters more than almost anything else. Give out too many and the price falls away, and with it the reason anyone would bother cutting emissions. The MSR was the change that stuck, swapping one-off political fixes for an automatic supply rule the market could actually rely on.
By 2026 the market is a settled, liquid one, with a price that is taken seriously, a surplus that is shrinking and a scope that keeps growing. The questions now are about ambition rather than survival: how quickly the cap should fall, how to manage the end of free allocation, and how the new ETS2 will behave once it starts trading. Part 3 picks up there.
Part 3 covers the 1.19bn 2026 cap, the 4.3% LRF, primary auctions on EEX, the secondary futures market, CBAM displacing free allocation, and the launch of ETS2.
Part 1 — EU ETS Primer · Part 3 — EU ETS in 2026 · Canadian Carbon Pricing Series
Primary sources for the phase-by-phase figures and the TNAC numbers cited above.
Climate Decode helps compliance buyers and project developers stay ahead of the policy curve across the EU ETS and beyond.