Carbon Accounting for Financial Services: Financed Emissions in Australia

Here's a number that should unsettle every risk team in Australian banking: financed emissions — the carbon embedded in lending and investment portfolios — are on average 750 times larger than a financial institution's direct operational footprint. In North America, that ratio hits 11,000 times. Your office electricity, company cars, and flights are a footnote. The real emissions sit in the loans you wrote.

And now Australia wants you to count them.

ASRS mandatory climate reporting is pulling financial institutions into Scope 3 Category 15 disclosure whether they're ready or not. Asset owners managing $5 billion or more in funds are ASRS Group 2 by default — reporting for financial years from 1 July 2026. The big four banks hit Group 1 thresholds a year earlier. Every APRA-regulated entity with significant lending or investment portfolios will eventually need to measure financed emissions in Australia, disclose them publicly, and defend the methodology under assurance.

We build carbon accounting technology, and we'll be honest: financed emissions are the single hardest problem in this field. Harder than tracking diesel across construction sites. Harder than collecting Scope 3 data from reluctant suppliers. Because you're not measuring what your own business does — you're measuring what thousands of borrowers and portfolio companies do, often with patchy data and incomplete reporting from their end.

This article is about the mechanics. How financed emissions actually get calculated, asset class by asset class. Where the data breaks down. What APRA and the AASB expect. And what a realistic first-year disclosure looks like when you can't get perfect data from every counterparty in your book.

Your Carbon Footprint Is Your Loan Book

For a manufacturer, carbon accounting starts with gas bills and diesel tanks. For a bank, it starts with the general ledger.

Every dollar you lend to a coal-fired power station, a cattle farmer, a property developer, or a logistics fleet carries an attributed share of that borrower's emissions. The same applies to every equity holding in your investment portfolio, every mortgage on your book, every infrastructure project you financed. Financed emissions account for over 99% of a typical financial institution's total greenhouse gas inventory. The other 1% — Scope 1 and 2 from your branches and data centres — barely registers.

This is Scope 3 Category 15 under the GHG Protocol. And it's the category that turns financial services carbon accounting from a facilities management exercise into a data infrastructure problem.

The methodology that's become the global standard is PCAF — the Partnership for Carbon Accounting Financials. Started by 14 Dutch banks in 2015, it now has over 700 signatories worldwide. In December 2025, PCAF released its third edition, expanding from seven to ten asset classes and adding methodologies for securitisations, use-of-proceeds structures, and sub-sovereign debt. Commonwealth Bank sits on the PCAF Global Core Team. The standard is the reference point for AASB S2 financed emissions disclosure.

Here's the core formula. It's simple in concept and brutal in execution:

Financed Emissions = Attribution Factor x Borrower/Investee Emissions

The attribution factor determines what share of a company's total emissions gets allocated to your institution, based on how much of that company's financing you provided. The borrower's emissions are their actual (or estimated) Scope 1, 2, and — where available — Scope 3.

But that formula hides enormous complexity. The attribution factor calculation differs for every asset class. The borrower emissions data ranges from audited CDP disclosures to pure guesswork based on sector averages. And you need to do this across potentially tens of thousands of exposures.

How the Calculation Works Across Asset Classes

PCAF defines methodologies for different types of financial assets. The attribution factor — the fraction of borrower emissions you claim as yours — works differently depending on the asset class. Here's how.

Listed Equity and Corporate Bonds. The attribution factor is your outstanding investment amount divided by the company's Enterprise Value Including Cash (EVIC). If you hold $50 million of shares in a mining company with a $10 billion EVIC, you attribute 0.5% of that company's emissions to yourself. The data is relatively available here because listed companies increasingly report emissions. But EVIC fluctuates daily with share price, which means your financed emissions can swing materially without a single physical change in the real world. That's a disclosure headache — do you use year-end EVIC, average EVIC, or point-in-time? PCAF recommends using the investee's most recently available EVIC.

Business Loans and Unlisted Equity. Same principle, but the denominator switches to the borrower's total equity plus debt at book value. If you extended a $20 million facility to a private company with $100 million in total balance sheet financing, you attribute 20% of their emissions. The data problem is worse here because private companies rarely report emissions. You're often working with ANZSIC sector averages and revenue proxies instead of actual data.

Mortgages. Attribution is based on the outstanding loan amount divided by the property value at origination. A $600,000 mortgage on an $800,000 home means you attribute 75% of that building's operational emissions. The emissions themselves come from estimated energy use based on property characteristics — floor area, building type, energy rating, location. Australia's residential building stock is wildly inconsistent, from NatHERS 2-star homes built in the 1970s to 7-star new builds. Getting reliable per-dwelling emissions without actual energy consumption data is a significant estimation challenge.

Project Finance. You attribute emissions based on your share of total project financing. If your bank provided $200 million of a $1 billion LNG project's total debt and equity, you attribute 20% of the project's emissions. The difference from business loans is that project finance has a known use of proceeds — you know exactly what the money built. That makes the emissions calculation more targeted but also more exposed to scrutiny, because the assets are identifiable.

Commercial Real Estate. Similar to mortgages but for income-producing property — offices, retail, industrial. Attribution is loan outstanding divided by property value. Emissions come from the building's operational energy use. This is one area where Australian data is improving, because NABERS ratings give you a direct energy intensity benchmark for office buildings. But not every commercial property has a NABERS rating, and the ones that don't tend to be the worst performers.

Motor Vehicle Loans. Attribution is loan outstanding divided by vehicle value. Emissions come from estimated annual kilometres driven and the vehicle's fuel consumption or electricity use. Fleet lending portfolios can use actual fuel card data if available, which pushes data quality up significantly.

The Data Quality Problem Is the Whole Problem

We've been building data extraction systems for carbon accounting for years, and financed emissions makes every other data problem look small. When you're calculating Scope 2 from electricity bills, at least you have the bill. When you're doing financed emissions, you often have nothing but a loan balance and an industry code.

PCAF's data quality scoring system acknowledges this honestly. It runs from Score 1 (best) to Score 5 (worst):

Score 1 — Verified, audited emissions reported directly by the borrower. Think: a listed company with a CDP A-list disclosure and third-party assurance on their emissions data. This is rare even for large corporates.

Score 2 — Unverified self-reported emissions from the borrower. They gave you numbers but nobody's checked them. Most ASX200 company data sits here.

Score 3 — You calculate emissions using the borrower's physical activity data (energy consumption, fuel use, production volumes) and apply emission factors yourself. This is where our kind of technology — AI-powered extraction from utility bills and operational data — actually helps, because you're turning primary data into emissions estimates.

Score 4 — Sector-average emissions intensity applied to the borrower's revenue or assets. You know the company is in the "road freight" sector and made $50 million in revenue, so you multiply by the sector's average emissions intensity per dollar of revenue. This is where most business loan portfolios sit today.

Score 5 — Pure proxy estimates with minimal data. You know the loan size and maybe the ANZSIC code, and that's it. This is where SME lending portfolios live, and it's where the numbers are least reliable.

Here's the uncomfortable truth about Australian financial services right now: for most institutions, the weighted average data quality score across their lending book is probably sitting between 3.5 and 4.5. The listed equity portfolio will score better (more reported data). The residential mortgage book and SME lending will score worse.

And that matters, because AASB S2 and the assurance framework don't just want a number — they want you to disclose your data quality. Your auditor will ask how much of your financed emissions total is based on verified data versus sector proxies. A disclosure that says "87% of our financed emissions are estimated using Score 4-5 data" is technically compliant but also a flashing signal to investors and regulators that the number could be materially wrong.

We're not sure there's a shortcut here. Improving data quality means either getting borrowers to report better data (slow, requires engagement programs, and many SMEs don't have the capacity) or getting access to their actual utility and operational data to calculate emissions on their behalf. That second approach is where we think the industry needs to go — and it's where carbon accounting software that can process primary source documents at scale becomes genuinely useful for financial institutions, not just for the borrowers themselves.

What APRA Actually Expects

APRA's Prudential Practice Guide CPG 229, finalised in November 2021, established the framework for how banks, insurers, and super funds should manage climate-related financial risks. It doesn't mandate a specific emissions measurement methodology. But it makes the direction unmistakably clear.

In APRA's 2024 Climate Risk Self-Assessment Survey — covering 149 APRA-regulated entities — 54% reported having set at least one climate-related target, with financed emissions among the most common basis for target-setting. That means over half of surveyed institutions are already committing to financed emissions targets before ASRS even requires them to disclose the numbers.

But APRA also found that metrics and targets, and disclosure, were the weakest areas of CPG 229 alignment across the sector. Governance and risk management scored better. The actual counting — the financed emissions measurement — lagged behind.

APRA isn't standing still. In 2025, they began consulting on amending Prudential Standards CPS 220 and SPS 220 to explicitly include climate risk. That moves climate from guidance (CPG 229 is a practice guide, not a binding standard) to prudential requirements. The message is that measuring financed emissions isn't optional for APRA-regulated entities, and the regulatory ratchet only tightens from here.

For superannuation funds specifically, the exposure is enormous. Australia's super system holds over $4 trillion in assets. AustralianSuper alone has committed to net zero portfolio emissions by 2050 and measures the carbon footprint of roughly 70% of its investment portfolio. UniSuper targets a 43% emissions reduction by 2030. These targets are meaningless without reliable financed emissions measurement, and the ASRS framework will require the numbers behind the promises.

The ASRS Layer: What Financial Institutions Must Disclose

AASB S2 requires entities engaged in financial activities — asset management, commercial banking, insurance — to disclose financed emissions as part of their Scope 3 Category 15 reporting. This isn't a suggestion. It's a specific requirement in the standard.

Here's the timeline that matters:

Asset owners with $5 billion or more in funds under management are carved into ASRS Group 2 — reporting for financial years beginning on or after 1 July 2026. They're explicitly excluded from Group 1, so even very large super funds and registered schemes begin reporting with Group 2. They get a one-year deferral on Scope 3, meaning financed emissions become mandatory in their second reporting year (FY28 for most).

The big four banks — CBA, NAB, ANZ, Westpac — easily meet Group 1 thresholds and are already in their first ASRS reporting cycle. Their Scope 3 deferral expires next year.

In December 2025, the AASB approved amendments (AASB S2025-1) that provide some practical relief for financial institutions:

You can limit Category 15 disclosure to financed emissions only — excluding derivatives, facilitated emissions, and insurance-associated emissions unless you choose to report them or your regulator requires it. That's a meaningful scope reduction.

You don't have to use GICS (the Global Industry Classification Standard) to disaggregate financed emissions by industry. You can use ANZSIC or another system that fits your portfolio better. For Australian banks, this is a practical win because ANZSIC is already embedded in lending systems.

The December 2025 amendments also allow jurisdictional relief on GHG measurement methods and global warming potential values. NGER uses AR5 GWP values; AASB S2 technically requires AR6. That mismatch creates real headaches, and the amendments give some flexibility here.

But here's what the amendments don't change: you still need to disclose absolute gross financed emissions, disaggregated by Scope 1, Scope 2, and Scope 3 of your borrowers and investees, broken out by industry and asset class. That's a massive data aggregation exercise. And the modified liability framework protects Scope 3 disclosures from private litigation for three years — but ASIC can still act, and after 31 December 2027, full liability applies.

A Realistic First-Year Approach

If you're a mid-market insurer, a regional bank, or a $5B+ super fund staring down your first financed emissions disclosure, here's what a realistic first year looks like. Not the aspirational version. The version that survives assurance.

Start with your listed equity and corporate bond portfolio. Data quality will be highest here. Most ASX200 companies report emissions. Many are verified. You'll get Score 1-2 data for a meaningful portion. Use EVIC-based attribution. This gives your assurance provider something solid to anchor on.

For business loans, accept that Score 4 is your starting point. Map your commercial lending book by ANZSIC code. Apply sector-average emissions intensity factors from PCAF's database or from published Australian data. Prioritise your top 20-30 exposures by loan value for direct engagement — ask them for actual emissions data or at least energy consumption data you can convert. That gets your biggest loans to Score 2-3 and moves the needle on your weighted data quality.

Mortgages are a volume game. You might have 100,000 home loans on your book. You won't get individual energy data for each one. Use state-based average residential emissions factors, adjusted for property type and size where you have the data. NatHERS ratings help where available. Accept Score 4-5 for the bulk of the book and disclose that transparently.

Disclose your data quality. PCAF requires it. Your auditor will ask for it. And honestly, transparent disclosure of Score 3-5 data quality in year one is far better than claiming precision you don't have. The modified liability framework protects good-faith estimates. What it doesn't protect is numbers that your board signed off on without understanding the methodology behind them.

Document your improvement plan. Show how you intend to move from Score 4-5 towards Score 2-3 over three to five years. Which sectors will you prioritise for engagement? What data infrastructure are you building? This is what APRA wants to see — not perfection, but a credible trajectory.

One thing we've learned from building carbon accounting tools: the institutions that start with honest, well-documented estimates in year one build the trust and the systems to produce better numbers in year three. The ones that wait for perfect data never report at all.

Where We Fit — and Where We Don't

We should be transparent about Carbonly's role here. We're good at extracting emissions data from utility bills, fuel invoices, and operational documents using AI. That capability maps well to Score 3 data quality — taking a borrower's actual energy consumption data and calculating emissions from it.

For a bank running a borrower engagement program, we can help your commercial clients get their Scope 1 and 2 numbers right faster and cheaper than hiring a consultant. That feeds directly into your financed emissions calculation as higher-quality input data.

But Carbonly also has capabilities that map directly to how financial institutions manage climate data internally. Our JV Collaboration module handles equity-based emission allocation — the same attribution logic that underpins financed emissions. For fund managers co-investing in assets alongside other institutions, this means you can model the shared ownership structure and allocate emissions by equity share, which mirrors the PCAF attribution factor methodology at the asset level. It doesn't replace a portfolio-level aggregation engine, but it gives you clean, auditable emission allocations for your directly held and co-invested assets.

Custom Dashboards let you build portfolio-level views with drag-and-drop widgets — emissions by sector, by asset class, by data quality score — and share them with investment committees or risk teams without giving everyone full platform access. Scheduled Reports mean your quarterly financed emissions summary can be automatically generated and delivered to the investment committee, the board risk committee, or APRA — on a set cadence, in the format they expect, without someone manually pulling data the week before the meeting.

For institutions setting net zero targets, Carbonly's Targets module supports SBTi alignment with baseline period configuration and progress tracking against science-based reduction pathways. Financial institutions face specific SBTi requirements — the SBTi Financial Institutions framework requires portfolio-level targets — and tracking progress against those targets needs a system that connects your emissions data to your commitment, not a slide deck updated once a year. Every change is logged in the Audit Trail, so when your assurance provider asks how a target was set, what the baseline was, and when a methodology changed, the answer is in the system — not in someone's email archive.

We don't build portfolio-level financed emissions aggregation engines today. That's a different product — one that sits on top of your lending system, pulls exposure data, matches it to borrower emissions (however sourced), applies attribution factors by asset class, and produces the disaggregated disclosure AASB S2 requires. The big four banks are building or buying those systems themselves. Mid-market institutions will need to decide whether to build, buy, or cobble together spreadsheets. (We know which option we'd pick. And we know which one most will actually choose.)

The honest gap in the market right now is at the borrower level. If you could get 500 of your largest commercial borrowers to accurately report their emissions, your financed emissions data quality would improve dramatically overnight. That's the piece we're working on — making it so easy for a borrower to produce an emissions number that there's no excuse not to.

The Greenwashing Risk Nobody Talks About

Here's a risk that sits outside the accounting methodology but is just as real: every financed emissions target your institution publishes is an ACCC exposure.

When CBA says they're reducing financed emissions in their fossil fuel portfolio, or when a super fund commits to net zero by 2050, those are environmental claims. The ACCC has levied over $42 million in greenwashing penalties in the past twelve months — including $12.9 million against Vanguard for misrepresenting ESG screening in its "Ethically Conscious" fund, and $10.5 million against Active Super for investing in securities it claimed were excluded.

If your financed emissions methodology is built on Score 5 estimates and you're publishing targets against those numbers, you're making claims you can't substantiate. That's exactly the pattern the ACCC is prosecuting.

The fix is boringly simple: match the precision of your claims to the quality of your data. If your weighted average data quality is Score 4, don't publish sector-level reduction targets with implied decimal-point precision. Say what you know, disclose what you estimated, and show how you're getting better.

Financed emissions aren't going away. APRA is moving climate into binding prudential standards. ASRS is making Category 15 disclosure mandatory. The PCAF methodology will keep evolving — the third edition's expansion to ten asset classes signals a standard that's becoming more detailed, not less.

The institutions that start measuring now — even imperfectly — will be the ones that can actually manage their climate risk exposure in three to five years. The ones that treat this as a compliance checkbox will keep being surprised when the numbers tell a story their board wasn't expecting.

If you're a bank or super fund and your borrowers need help getting their emissions data right, that's a conversation we should have. Not because we'll solve your financed emissions problem overnight — nobody can. But because the quality of your financed emissions disclosure is only ever as good as the data your borrowers give you. And right now, most of them are giving you nothing.

Related Reading: