2026年4月、GX-ETSは義務化された——実際に変わったこと
今年4月、日本の排出量取引制度は任意ではなくなった。基準は年間10万トン、未達時の負担は上限価格の1.1倍、オークションの開始は2033年度。この設計が、東京が何を最適化しているかを語っている。
For three years, joining Japan’s carbon market was a reputational decision. A company signed up to the GX League, set its own reduction target, and traded if it wanted to. More than 700 companies did, together accounting for over half of Japan’s national greenhouse gas emissions (ICAP).
On 1 April 2026 that stopped being optional. The GX-ETS entered its mandatory phase, and for a defined set of large emitters the surrender of allowances became a legal obligation rather than a pledge (GX Acceleration Agency, the government body that will administer the scheme’s auctions and levy).
The interesting part is not that Japan now has mandatory carbon pricing. It is what the design was optimised for — and the answer is not price.
What changed
The scheme covers entities with annual CO₂ emissions of 100,000 tonnes or more, in power, industry, transport and aviation, and in any other sector that crosses the same threshold (ICAP). The legal basis is the amended GX Promotion Act, adopted by the Upper House of the Diet in May 2025 (ICAP), with the operational phase beginning in fiscal 2026 (GX Acceleration Agency).
Covered entities surrender one compliance unit per tonne of CO₂, and reported emissions require third-party verification by an accredited verifier. Upper and lower price limits apply to the traded allowance.
If an entity cannot surrender the required allowances, it owes a payment equal to the shortfall multiplied by the upper price, multiplied by 1.1 (ICAP). Offset credits are permitted — domestic J-Credits and international JCM credits — but capped at 10% of the compliance obligation.
Baseline-and-credit, not cap-and-trade
The design detail worth pausing on is the mechanism. The EU ETS is a cap-and-trade system: a fixed quantity of allowances is issued, and the market discovers the price. Japan built a baseline-and-credit system with a price corridor (ICAP).
The difference is where the certainty sits. Cap-and-trade fixes the environmental outcome and lets the cost float. A baseline-and-credit scheme with upper and lower price limits does closer to the reverse — it puts a band around what carbon will cost a Japanese emitter and accepts more variance in the aggregate tonnage.
Read alongside a shortfall payment set at the ceiling price plus ten percent, the intent becomes legible. A penalty pinned to the ceiling gives a company a computable worst case. It is a compliance regime a CFO can budget for, and it was built that way deliberately.
The runway is long, and that is the signal
Three dates define the ramp.
FY2026 — mandatory surrender begins, with allowances allocated rather than sold.
FY2028 — the GX Surcharge starts: a levy on fossil fuel importers and domestic extractors, priced on the CO₂ content of the fuel, collected by the GX Acceleration Agency (ICAP; GX Acceleration Agency).
FY2033 — auctioning is introduced, and only for high-emitting corporations in the power sector (ICAP; GX Acceleration Agency).
Seven fiscal years between the obligation and the first auction. For most of that period, the allowance is something a covered company receives rather than buys. Layer that on the instruments already in place — a Tax for Climate Change Mitigation levied on fossil fuels since 2012, and sub-national emissions trading systems in Tokyo and Saitama (ICAP) — and the shape of the policy is a slow tightening across three separate instruments, none of which delivers a sharp price shock on its own.
Whether that is prudence or delay is the live argument. What it is not is accidental. Japan has stacked its carbon pricing so that the cost signal arrives after the measurement infrastructure does.
The binding constraint is verification, not price
Here is the practical consequence for a company inside the threshold.
Reported emissions require third-party verification by an accredited verifier, every year, against an obligation with a computable financial penalty attached. Before a single yen of carbon cost lands, a covered emitter needs an emissions inventory that survives an external check — consistent boundaries, documented methodology, source data that can be traced back through the operating systems that produced it.
Most large Japanese emitters have been reporting emissions for years, under the Act on Promotion of Global Warming Countermeasures and to whichever voluntary frameworks their investors asked about. Those numbers were produced for disclosure. Numbers produced for disclosure and numbers produced for a surrender obligation are held to different standards, because only the second kind is arithmetic that money is calculated from. The gap between the two is where the next three years of work sits — in data lineage, in verifier capacity, and in the internal controls that let a Japanese company assert a tonnage figure the way it asserts a revenue figure.
This is also where the same problem shows up from the other direction. Japanese exporters selling into the EU are already inside the CBAM’s definitive phase, which requires verified embedded-emissions data at the product level. A domestic scheme demanding entity-level verified data and a trade regime demanding product-level verified data are two obligations that can share one evidence chain, if a company builds it that way — and two parallel reporting programmes if it does not.
Who this moves
Industrial emitters near the line. The 100,000-tonne threshold is a cliff edge, not a slope. Companies sitting close to it have a boundary question — which facilities, which entities, which fiscal years — that is worth resolving before it is resolved for them.
Measurement and verification providers. Every covered entity needs accredited third-party verification annually. That is a capacity constraint on a market with a legislated start date, and capacity constraints on regulated deadlines are where climate-tech businesses get built.
Investors in Japanese industrials. From FY2028 the surcharge reaches fuel importers, and from FY2033 power generators start bidding for allowances. Neither is priced into a 2026 model that treats Japanese carbon cost as approximately zero.
Three questions worth putting to a room
If the people who set these rules and the people who have to live with them are in the same place, the useful questions are narrow ones.
What happens to a covered emitter that is structurally short — an industry where the abatement technology is not commercial before 2033, and whose only compliance path is to pay the ceiling plus ten percent every year? A penalty that is computable is also a penalty that can be budgeted as a permanent line item, which is not the behaviour the scheme wants.
Is verifier capacity being built ahead of the FY2028 and FY2033 steps, or will it be procured in the year it is needed? Verification is the load-bearing element of the whole design, and it is the element with the longest lead time.
And can one verified emissions chain serve both the GX-ETS and CBAM, or will Japanese exporters run two? The answer is mostly an accounting and systems question today, and it becomes a competitiveness question by 2028.
None of these have settled answers, which is what makes them worth an hour with the people who will decide them.
Tech for Impact Summit 2027 takes place in Tokyo on 18–19 May 2027, as a partner event of SusHi Tech Tokyo. Climate and energy is one of its core tracks, and the summit is invitation-only — the room is built by curating who is in it. If your work sits on this question, whether as a covered emitter, a verifier, a policymaker or the capital behind any of them, you can ask to be considered at tech4impactsummit.com/apply.