GX-ETSの無償割当はどう決まるのか：二つの算定式と、すでに終わった基準期間

Japan’s GX-ETS covers companies with direct CO2 emissions over 100,000 tonnes, measured as a three-year average of FY2023, FY2024 and FY2025. The same three years decide how many free allowances each of those companies receives.

Two formulas turn that history into an allocation. Which one governs a given process is what decides the cost.

Every covered company knows the obligation: surrender allowances against verified emissions. Fewer have worked out that their free allocation was computed from the output and emissions they reported for FY2023 to FY2025.

Your exposure is the gap between emissions and allocation

A carbon market prices the gap between what you emit and what you were given. Under the GX-ETS, allowances in the mandatory phase are allocated rather than auctioned, with auctioning introduced for high-emitting power companies from FY2033 (ICAP).

For the entities covered now, the obligation side is set by operations and the allocation side by a government formula. The exposure is the difference. Two companies with identical emissions can face very different costs depending on which formula governs their processes and how their baseline years looked.

The Ministry of Economy, Trade and Industry has published the design in its Industrial Structure Council subcommittee materials (METI). The inputs are historical and cannot be changed now, which is reason enough to read it closely.

Benchmarking pays against the top performers in your industry

Benchmarking applies, in METI’s words, “primarily for energy-intensive sectors,” where industry characteristics matter enough to justify a sector-specific calculation.

The method sets a performance standard rather than grandfathering a company’s own history. METI describes it this way: “For each industry, the benchmark level is determined by comparing companies’ emissions intensity per unit of output. The benchmark is set at the performance level of the top X% within the industry.”

The allocation is then:

Allocation = Baseline activity level × Target emission intensity for each year

The baseline activity level is “average production, etc. over the three years before the introduction of the ETS, i.e. FY2023 ~ FY2025.”

The standard is relative to peers rather than to your own past, so an efficient plant in an inefficient industry is rewarded and the reverse is penalised. METI’s published diagram also shows the benchmark level being “gradually lowered each year,” so free allocation per unit of output tightens across the phase even when output is flat.

The intensity value for each industry is not in the published design. METI has set up working groups to build the formulas, noting that “benchmark formulas require industry-specific expertise.” Industries are selected on two criteria: “the degree of institutional impact (emissions by industry, the number of operators, impact on industrial policy)” and “technical difficulty (comparability between businesses based on the complexity of manufacturing processes, diversity of varieties, etc.).”

A Manufacturing BM Study WG covers petroleum refining, steel, chemicals (petrochemical, soda, carbon black), pulp and paper, cement, lime production, aluminum, rubber products, flat glass, glass bottles and automobiles. A separate working group covers power generation. Transport sits with the ministries that regulate it, each reporting its own benchmark proposal back to the subcommittee.

ICAP’s factsheet lists the benchmarked activities in the same shape and adds freight road transport, domestic marine transport and air transport alongside the manufacturing processes.

Grandfathering cuts a fixed percentage from your own history

Not every sector can be benchmarked. Where products are too varied or processes too dissimilar for a per-unit comparison to be fair, METI applies grandfathering, “which applies a certain reduction rate to the emissions of a baseline year.”

The allocation is:

Allocation = Baseline emissions × (1 − Target reduction rate × years since the baseline)

Baseline emissions are the “average emissions over the three years before the introduction of the ETS i.e. FY2023~FY2025.”

This is the simpler formula and the less forgiving one over time. The subtraction is linear in years elapsed, so the allocation declines on a fixed schedule from a fixed starting point. A company that did its cheap abatement before FY2023 locked in a lower baseline, then faces the same percentage decline from it. Early movers are penalised by the arithmetic. It is a known weakness of grandfathering in every system that uses it, and Japan has not exempted itself from it.

One company gets both formulas, process by process

The two formulas are not alternatives at company level. They are applied per process.

METI states that an entity’s total allowance “will be calculated by summing the allowances for each benchmark-covered process and each grandfathering-covered process at each of the entity’s facilities, and then applying adjustments based on relevant consideration factors.”

For a diversified industrial group that is the demanding part. Emissions and activity data have to be resolved to the process level at each facility, mapped to whichever method governs that process, for three historical years. An inventory built to report one company-wide total does not answer that question.

The baseline is FY2023 to FY2025, because that data already existed

Both formulas depend on the same window, and METI’s reasoning for picking it is published.

Averaging exists to damp volatility: emissions and activity levels “are affected by fluctuations in economic activity and other factors and therefore may vary by fiscal year.” Longer windows damp more and cost more to assemble. METI notes that “in other countries’ ETS systems, a 3-to-5-year average is commonly used,” and that “extending the calculation period increases the data collection burden on companies.”

The deciding factor was data that already existed: “Considering that many businesses have already calculated and verified their emissions during Phase 1 of the GX-ETS (FY2023–2025), the baseline period under the new system will be set as the three fiscal years immediately preceding participation in the system (FY2023–2025 for entities joining in FY2026).”

The same three years decide coverage. The scheme applies to “Corporations with direct CO2 emissions of over 100,000 tons,” measured as a “3-year-average from FY2023 to FY2025.” METI’s comparison table puts the EU ETS on a five-year average or median and Korea’s K-ETS on a three-year average calculated from the three years before each compliance period.

A company above 100,000 tonnes on that three-year average is in the scheme, and the same three years fix the baseline it is allocated against. Both numbers come from inventories it built and had verified while participation was still voluntary.

A closed plant is excluded and an inspection year is carved out

METI acknowledges that a flat three-year average “could lead to either excessive or insufficient allocations” where something structural happened in the window, and names three cases: a facility established during the period, a facility closed during the period, and reduced operation “due to unavoidable circumstances such as disasters or mandatory inspections.”

The published treatment is specific. For newly established facilities, “the baseline shall be calculated as the average of the period since establishment.” Facilities closed during the baseline period “shall be excluded from allocation.” For disasters or mandatory inspections, baseline figures are “calculated based on the average excluding the affected period.”

A plant idled for a periodic inspection in FY2024, or a line commissioned in FY2025, moves the arithmetic. The adjustment is claimed rather than granted, and the evidence for it is two and three years old.

A chemicals group runs three allocation calculations at once

Picture a chemicals group running a carbon black line, a soda plant and its own freight fleet. Carbon black and soda both sit on METI’s manufacturing benchmark list, so each gets a benchmark intensity multiplied by its FY2023 to FY2025 average output. The freight operation is benchmarked by the transport ministry on its own formula. Anything else the group emits falls to grandfathering, against its own three-year average emissions.

That is three separate allocation calculations inside one corporate boundary, and the data has to come out at process level by facility. A group that has only ever reported one company-wide tonnage does not have it in that shape.

The same figures face an accredited verifier, because reported emissions require third-party verification and the allocation is drawn from those records. If the soda plant was down for a mandatory inspection in FY2024, the exclusion is available, and the maintenance records that prove it are two years old.

METI has not yet published the benchmark intensity for carbon black or soda. Until the working groups report, that figure is the only input the group can still influence, and its industry association is the route in.

Korea and the EU anchor their baselines the same way

Korea’s K-ETS runs on a three-year average too, recalculated before each compliance period. The EU uses five years, or a median. Japan took its three years from a voluntary phase it had just finished running.

Take an electric-furnace steelmaker, one of the processes METI has sent to the Manufacturing BM Study WG. Its FY2023 to FY2025 output and intensity are now fixed inputs to a benchmark formula that is still being written. It can argue about the formula through its industry association. It cannot go back and change the years.

Any government moving from voluntary to mandatory climate reporting is choosing a baseline the same way. Companies reporting voluntarily in those jurisdictions today are filing numbers that may later price them.

Tech for Impact Summit 2027 runs on 18 and 19 May 2027 at the United Nations University in Tokyo, co-hosted by UNU. Senior leaders from business, policy and capital spend two days on questions of this kind. Participation is by invitation. If this is the work you do, ask us for one at tech4impactsummit.com/apply.