The $75 Trillion Data Problem Sitting on Bank Balance Sheets

Financed emissions are the greenhouse gases a bank owns through its loans and investments. Banks representing more than $75 trillion in assets have adopted the accounting standard for them. Only about a third meet its data-quality bar. That gap used to be a reporting footnote. In 2026 it is becoming a capital and collateral problem.

For years, financed emissions were treated as a disclosure exercise, a number for the sustainability report, produced once a year and rarely questioned. That framing is ending. Supervisors are turning the number into a risk input, and the moment they do, the quality of the data behind it stops being an ESG concern and becomes a treasury one. Just as climate disclosure has become a board-level liability, financed-emissions data is becoming a capital-level one. This is written for the people who manage capital and collateral, not the people who write the ESG report.

Financed emissions, in one line

Financed emissions are the emissions attributable to what a bank lends to and invests in. If you finance a steelmaker, a share of that mill's emissions is yours. The standard for measuring them is PCAF, the Partnership for Carbon Accounting Financials, and its reach is now enormous: more than 700 financial institutions have signed on, and firms representing over $75 trillion in assets have adopted the methodology. The Net-Zero Banking Alliance alone counts more than 330 banks holding upward of $100 trillion.

The number matters because it is large. For most banks, financed emissions dwarf their own operational footprint, often by hundreds of times. A bank can run its offices on renewable power and still own an enormous carbon position through its loan book. That position is what regulators have started to look at.

From a disclosure line to a capital input

Three moves are turning financed emissions from something a bank reports into something a bank is measured on.

First, the European Central Bank has folded climate into Pillar 2, the bank-specific layer of capital supervision, through the Supervisory Review and Evaluation Process (SREP). A bank that cannot show it understands and manages its climate risk can see that reflected in supervisory expectations.

Second, and more concretely, on 15 June 2026 the ECB introduced a climate factor into its collateral framework. When a bank pledges corporate bonds to borrow from the central bank, bonds from issuers more exposed to climate-transition risk now carry an additional haircut, so the bank can raise less liquidity against the same paper. The calibration draws on the Eurosystem 2024 climate stress test, the issuer's climate score, and the bond's maturity.

Third, the Basel Committee published its framework for climate-risk disclosure in June 2025, which includes financed emissions for material sectors. It is voluntary at the global level and becomes binding where national supervisors adopt it, which is the usual path from Basel guidance to hard rule.

How much this bites today, honestly

Not much, yet. The ECB itself expects the immediate impact of the collateral climate factor to be limited, because banks are borrowing modestly and corporate bonds are a small slice of pledged collateral. That is the wrong thing to focus on. The point is the precedent: a central bank now formally prices climate risk into how much a bank can borrow. Precedents in bank supervision do not shrink. They widen.

The data-quality gap nobody underwrote

Here is the part that should worry a risk officer. Adoption is near-universal among large banks. Data quality is not.

Out of roughly 300 banks that reported financed emissions for FY2024, only about 34% met PCAF guidance on disclosing their borrowers' Scope 1, 2, and 3 emissions with the required clarity. The rest reported numbers built on estimates, sector averages, and proxies. So the industry has committed $75 trillion of balance sheet to a metric that, two times out of three, it cannot yet stand behind with verified data.

That was tolerable when the number lived in a voluntary report. It is not tolerable when a supervisor can challenge the number and the capital held against it.

Why an estimate becomes a liability

PCAF scores every data point on a five-point quality scale, and the scale is the whole story. A score of 1 means the borrower's emissions were calculated and verified by a third-party auditor. A score of 2 means they were reported and calculated under the GHG Protocol, but not independently verified. Scores of 4 and 5 mean the bank used sector and region averages, or broad proxies, because real data was not available. Score 5 is a guess dressed as a figure.

A loan book weighted toward 4s and 5s is fine for a first voluntary disclosure. Under supervision it is exposure, for a simple reason: a number you cannot defend is a number a supervisor can move, and if the supervisor moves it up, the capital or the collateral treatment moves with it. Weak data does not just misstate your footprint. It misprices your book.

The Financed-Emissions Data Ladder

Reframe the PCAF scale as a capital-risk ladder rather than an accounting one. At the bottom sit scores 4 and 5: proxy and estimate. These are cheap to produce and carry the most supervisory and pricing risk, because there is nothing underneath them to defend. At the top sit scores 1 and 2: reported and verified. These cost more to produce and are defensible when challenged. Climbing the ladder is not an ESG project. It is a capital-planning move, and it only has to happen where it matters.

Most of a bank's financed emissions come from a handful of carbon-intensive sectors: power, oil and gas, cement, steel, transport. You do not need verified data on every counterparty. You need it on the sectors that dominate the number and would move your capital position if a supervisor pushed back. Concentrate the verification there.

What to do now

Three moves, in order.

Map your book's PCAF score distribution before a supervisor asks for it. If you cannot say what share of your financed emissions sits at score 4 or 5, that is the first finding.

Build verified measurement, reporting, and verification feeds for the two or three sectors that drive the number, moving those exposures up the ladder from proxy to reported to verified. This is the MRV work across energy and carbon markets that turns a soft estimate into a defensible figure.

Put data quality on the capital-planning agenda, not only the sustainability one. The question is no longer whether you have disclosed. It is whether you can defend the number if it is challenged, and what it costs you if you cannot. That reframing is the core of how we think about data integrity.

It is not only banks

Everything above applies to banks first because supervisors moved on banks first. It does not stop there. Insurers face the same accounting through PCAF's insurance-associated emissions standard, and their underwriting and investment books carry the same data-quality problem. Asset managers report the financed emissions of the portfolios they run, and inherit the data quality of every holding they own. And private credit, the fastest-growing pool of capital in the room, often has the thinnest data of all, because it lends to private companies that disclose the least.

The pattern is identical across all of them. Adopting the metric is easy and nearly done. Building defensible data underneath it is the hard part, and it is unevenly distributed. Whoever holds the most carbon-intensive exposure with the weakest data carries the most unpriced risk. For a private-credit or blended-finance provider, that is worth checking before a limited partner or a supervisor checks it for you.

The next two years

The pressure will arrive as a sequence, not a shock. Expect climate considerations to show up more explicitly in Pillar 2 outcomes. Expect the collateral climate factor to widen in scope and bite as corporate-bond pledging recovers. Expect more national supervisors to make the Basel climate framework mandatory. None of these is dramatic on its own. Together they move financed-emissions data quality from a reporting metric into a driver of the cost of capital by 2027 to 2028.

The banks that come through this well will not be the ones that disclosed earliest. They will be the ones that could defend their numbers when a supervisor finally asked, because they built verified data where it mattered before they had to. Everyone has already agreed to measure financed emissions. The advantage now goes to whoever can prove them.

Frequently asked questions

What are financed emissions?

They are the greenhouse gas emissions a bank or investor is responsible for through its loans and investments. If a bank finances a carbon-intensive company, a share of that company's emissions is attributed to the bank. For most banks, financed emissions are far larger than their own operational emissions.

What is PCAF and its data-quality score?

PCAF, the Partnership for Carbon Accounting Financials, is the standard for measuring financed emissions. It scores each data point from 1 to 5, where 1 is third-party verified data and 5 is an estimate based on sector or region averages. More than 700 institutions have adopted it.

Are financed emissions a capital requirement yet?

Not as an explicit, standalone capital charge. But climate risk is already inside Pillar 2 supervision at the ECB, and since 15 June 2026 the ECB collateral framework applies an extra haircut to climate-exposed corporate bonds. The direction is clear: financed-emissions data is moving toward capital and collateral treatment.

What is the ECB climate factor?

It is an additional haircut the ECB applies to certain corporate bonds pledged as collateral, based on the issuer's exposure to climate-transition risk. A more exposed issuer means a bank can borrow less against the same bond. It took effect in 2026 and is expected to widen over time.

How can a bank improve financed-emissions data quality?

By moving its most material exposures up the PCAF scale, from proxy and estimate (scores 4-5) to reported and verified (scores 1-2). The practical route is verified MRV on the carbon-intensive sectors that dominate the number, rather than trying to fix every counterparty at once.