Modelling the Safeguard Mechanism 2027 Baseline Decline Against a Live Ledger

Safeguard facility managers have had the same problem since the March 2023 reforms. They know the baseline is declining at 4.9% every year to 30 June 2030. They can quote the number in their sleep. What they cannot tell you, without pulling out a spreadsheet and locking themselves in a room for a week, is whether their facility is going to breach in FY28 or FY29, and by how much.

That is not a knowledge problem. It is a modelling problem. The Safeguard Mechanism forces a trajectory decision every year, and the decision has to be made against emissions data that is still landing. A spreadsheet snapshot taken in July tells you nothing about where you will be in June.

The reforms proposed for 2027 would make this harder. Tightening the decline mechanics from 1 July 2027 has been consulted on rather than legislated, and as at September 2026 none of it is settled law. On our reading of the published proposals, facilities that sat comfortably under baseline in FY26 could start to bump against it in FY28, and facilities already close could go over.

If you run a Safeguard facility, the next 12 months are when the modelling has to move off the spreadsheet and onto the live ledger.

What is actually declining, and by how much

The Safeguard Mechanism covers facilities with covered emissions above 100,000 tCO2-e per year. Since 1 July 2023, baselines have been declining 4.9% every year, on the way to a 30% reduction from 2020-21 levels by 30 June 2030.

The mechanics that matter for modelling:

Production-adjusted baselines. Most Safeguard facilities are on production-adjusted baselines rather than fixed baselines. That means the baseline itself moves with production volume, using a facility-specific emissions intensity (t CO2-e per tonne of product) that also declines at 4.9%.

Site-specific emissions intensity values (SSEIVs). New facilities and significant expansions can apply for a site-specific baseline. These carry their own decline schedules and are not automatic.

Safeguard Mechanism Credits (SMCs). Facilities that come in below baseline earn SMCs. Facilities above baseline surrender either SMCs or ACCUs. This creates the marketplace that most facility managers now have to model on the buy side, not the sell side.

Excess emissions and the multi-year monitoring period (MYMP). A facility that breaches in a single year has options. It can surrender ACCUs. It can apply to average over a two-year or three-year MYMP. It can enter a formal exceedance report. What it cannot do is ignore it.

The Clean Energy Regulator's Safeguard Mechanism overview at cer.gov.au/schemes/safeguard-mechanism has the current legislative position. The 2027 reform package sits on top of this. It is a consultation position rather than settled law as at September 2026, and if adopted as consulted on it would tighten the decline further from 1 July 2027.

The trajectory question every facility has to answer

The question is deceptively simple: given my current emissions profile, when do I breach?

Answering it requires four inputs and one calculation.

Input 1: The current baseline. This changes annually. For most facilities on production-adjusted baselines, it is the SSEIV multiplied by production volume.

Input 2: The baseline decline schedule. 4.9% per year to FY30, then whatever the post-2030 mechanism looks like (this is still being consulted on).

Input 3: Live emissions data. Not last year's NGER submission. Not the estimate from your annual budget. The current-year emissions ledger, kept up to date monthly, reconciled to source documents.

Input 4: Production forecast. The volume of the covered product you expect to make over the next three to five years. This is the piece that finance owns and that sustainability rarely has direct visibility into.

The calculation: modelled emissions against baseline, year by year, out to at least FY30. The gap in each year is your compliance position. Positive gap (below baseline) earns SMCs. Negative gap costs money.

Doing this once, on a spreadsheet, in July, tells you nothing you can act on. Doing it monthly, on a live ledger with the current-year data landing as it happens, tells you exactly when to act and how much to spend.

That is the difference between a Safeguard trajectory check that runs against the ledger and the same modelling done once at year-end. Plenty of consultants run this analysis well. The constraint is usually the data they are handed, which is a year-end extract rather than a live position. We wrote about the mechanics of regular trajectory checking in Safeguard Mechanism trajectory check and baseline breach prevention.

Which facilities are at risk from 1 July 2027

The reform package proposed to take effect from 1 July 2027 would do three things that matter for trajectory modelling. Each is a consultation position as at September 2026, not law:

• Baselines would tighten further beyond the current 4.9% annual decline for certain sectors.
• The SMC market rules would adjust, changing the effective cost of exceedance.
• Methane-specific management requirements would sharpen, particularly for gas processing, coal, and waste.

If the package lands broadly as consulted on, our reading is that the facilities most exposed to a FY28 breach are:

• Coal seam gas and LNG facilities where fugitives sit close to the current baseline.
• Coal mines where methane management is not already at Method 3 or 4 under NGER.
• Cement, lime, and steel where process emissions are structural rather than combustion-based.
• Waste facilities under the methane management plan regime.

The pattern that matters, on that reading: a facility currently within 5 to 10 percentage points of its baseline in FY26 is likely to be under pressure by FY28. That is not a huge margin. The 4.9% decline eats 5% of your headroom every year on its own, before production or reform changes.

The SMC-versus-abatement question

Once you have a modelled trajectory that shows a breach in year X, you have three decisions to make.

Decision 1: Buy SMCs or ACCUs. This is the market solution. SMCs and ACCUs trade at published prices. Multiply the shortfall (in tCO2-e) by the market price. That is your annual compliance cost if you do nothing.

Decision 2: Invest in abatement capex. This is the operational solution. Identify capex projects that reduce emissions at your facility. Calculate the marginal abatement cost (dollars per tonne avoided). Rank them from cheapest to most expensive. Draw the curve.

Decision 3: Some combination. For most facilities, the answer is a mix. Cheap abatement projects (below the ACCU price) are worth doing. Expensive abatement projects are cheaper avoided by buying credits. The MACC curve tells you the split.

This is exactly the work a MACC (marginal abatement cost curve) is designed for. Draw the curve. Overlay it with the current ACCU or SMC price. Everything to the left of the price line is worth doing. Everything to the right is worth avoiding. The MACC does not tell you what to do; it tells you where the crossover is.

MACC curves belong inside the same system as the emissions ledger, because the input to the MACC is the ledger data. Doing it in a separate spreadsheet means every time the ledger updates, the MACC is out of date.

What "against a live ledger" actually means

The phrase gets used loosely. Here is the concrete difference.

Snapshot modelling. You pull emissions data at a point in time (usually year-end), model the trajectory, present it to the exec committee, and file it. Three months later, production ramps up, or a Scope 1 factor gets updated, or a new anomaly gets caught. The snapshot is now wrong. Nobody re-runs it.

Live-ledger modelling. The trajectory is re-run against the live emissions ledger whenever new data lands or a period closes, so the exec committee is looking at the current position, not the position from three months ago.

The mechanics that make live-ledger modelling work:

1. Emissions data lands automatically from source documents (fuel dockets, gas meter exports, electricity invoices) through document ingestion, not manual entry.
2. Factor versions are pinned per period, so historical data does not shift when NGA factors update.
3. Anomalies are flagged as they land, not at quarter-end, so the ledger stays clean.
4. The MACC and the trajectory model sit against the ledger, not against exports of it.
5. The whole thing is audit-trailed so you can show what the position was on any given date.

Two of these deserve their own note. Factor version pinning is what stops your baseline year moving when DCCEEW updates NGA factors. We covered this in emission factor versioning and the audit trail. If your baseline year emissions are re-calculated at every factor update, your Safeguard modelling becomes untrustworthy.

Second, the AR5 versus AR6 GWP wrinkle applies here too. NGER (which is what your Safeguard baseline is calculated against) uses AR5. AASB S2 disclosure can use AR6. Your Safeguard trajectory has to be modelled on AR5 to be consistent with the baseline. If you are modelling on AR6, your numbers will not line up with what CER assesses you against. AR5/AR6 GWPs and their calculation impact walks through the mechanics.

What to do in the next 90 days if you are a Safeguard facility

Concrete steps, in order.

Step 1: Confirm the baseline for FY27. Your production-adjusted baseline for FY27 has already been set (as of 1 July 2026). Confirm the SSEIV, confirm the applied decline factor, and confirm the production assumption. Sign it off in writing. This is the reference point for every model that follows.

Step 2: Reconcile FY26 emissions to closure. Your FY26 NGER submission is due by 2 November 2026, because 31 October falls on a Saturday. Whatever you submit is what CER uses to assess your FY26 position. Reconcile it before you submit. Do not submit and then find the reconciliation problem.

Step 3: Model the FY27 to FY30 trajectory. With FY26 closed, forecast FY27 through FY30 emissions on your best production assumption. Overlay the declining baseline. Identify the crossover year. That is your breach year.

Step 4: Cost the compliance gap in each scenario. Calculate the shortfall in tonnes for each year. Multiply by the current ACCU price (indicatively in the mid-A$30 to low-A$40 range as at September 2026, so check the spot price before you rely on it) and by CER's SMC price. This gives you a range for the annual compliance cost of doing nothing.

Step 5: Build the MACC. List every abatement project your engineering team can nominate. Cost each one. Estimate the annual abatement in tonnes. Divide capex plus opex-per-year by tonnes-abated-per-year. That is the marginal abatement cost. Rank them. Overlay the ACCU price. Everything left of the line is worth doing on cost grounds alone.

Step 6: Bring it to the exec committee. The paper needs three things. The trajectory. The compliance cost range. The MACC. The decision at exec level is which projects go into the capex plan and which are covered by credits.

Step 7: Set up regular re-modelling. This is the piece that separates surviving from thriving. The trajectory model has to re-run every month against the live ledger. Not because the decision changes every month, but because the position does. When production ramps, when a new emissions source is discovered, when an anomaly is caught, the model has to update.

Where the modelling breaks

We should be honest about what is hard here.

Production forecast uncertainty. For most facilities, the biggest single source of trajectory uncertainty is not emissions intensity. It is production volume. If your production forecast is soft (mining, oil and gas, cement), your trajectory is soft. Scenario ranges rather than point estimates are the right answer.

Reform package finalisation. The 2027 reform package is still being consulted on. Some of the specifics (particularly around methane-specific management for coal, and the SMC market rules) are not fully final. Modelling has to include a sensitivity for reform outcomes.

Facility-level data granularity. Safeguard is assessed at the facility level. If your emissions ledger is organised at the corporate group level rather than the facility level, the trajectory modelling gets messy fast. This is a data structure problem more than a modelling problem. Fix the structure first.

Interaction with AASB S2 disclosure. Your Safeguard position (facility level, AR5) and your AASB S2 disclosure (corporate level, AR6 where you choose) are related but not the same. Reconciling them is a piece of work that most reporters underestimate. The materiality threshold on emissions errors is worth reading on the assurance side of this.

What we would push a Safeguard facility to do this quarter

If we were sitting inside a Safeguard facility this July, three things would go on the top of the list.

Move the trajectory model onto the live ledger. Not next year. Now. Every month that the model sits in a spreadsheet is a month where the exec committee is making decisions on stale data.

Get the MACC built and reviewed. Even if the capex plan does not need it yet, the MACC forces the conversation between engineering, finance, and sustainability that produces the eventual investment case. Building one properly typically takes a few weeks of engineering and finance time. Start now.

Book a Q3 exec review specifically on Safeguard trajectory. Not lumped into a general climate paper. A specific hour on where the facility sits, what the FY28 breach risk looks like, and what the compliance cost range is. Once a year is not enough. Twice a year is minimum. Quarterly is right.

To see your own facility's breach year against a live ledger, bring a recent NGER submission or your last Safeguard baseline determination to hello@carbonly.ai. Carbonly's Safeguard Trajectory Check compares projected emissions with each facility's declining baseline, run on demand against the ledger.