Integrating Climate Risk into IFRS 9 Expected Credit Loss Models

Climate change has become a mainstream credit risk, and supervisors increasingly expect it to be visible in the numbers that banks report. Under IFRS 9, expected credit losses (ECL) must reflect reasonable and supportable forward-looking information, and both physical risk (floods, heat, storms damaging collateral and cash flows) and transition risk (policy, technology and market shifts that erode the value of carbon-intensive assets) clearly qualify. Yet for most banks these risks still sit outside the core models, captured through post-model management overlays rather than in the machinery that produces the estimate.

That gap is now the focus of regulatory attention. The direction of travel from Basel, the European Central Bank (ECB) and the UK Prudential Regulation Authority (PRA) is consistent: climate risk should be measured with the same rigour as any other driver of credit loss and, over time, moved from ad hoc adjustments into risk-sensitive, model-based estimation. This article sets out why that expectation is hardening, the practical methods banks are using, the data obstacles, and the disclosure consequences under IFRS 7.

Why Basel, the ECB and the PRA now expect climate in ECL

The regulatory foundations are not new. The Basel Committee's guidance on credit risk and accounting for expected credit losses has long required banks to consider all reasonable and supportable information, including forward-looking macroeconomic factors, when measuring ECL. Climate is simply the newest and least well-understood of those factors. What has changed is the intensity of supervisory follow-through. In July 2024 the ECB published its report on IFRS 9 overlays and model improvements for novel risks, drawing on granular AnaCredit exposure data and pressing banks to make provisioning more risk-sensitive. In the United Kingdom, the PRA finalised Supervisory Statement SS5/25 in December 2025, replacing the earlier SS3/19 and effectively putting climate risk management on the same footing as other financial risks. Firms are expected to complete a gap analysis against SS5/25 by 3 June 2026, with supervisory engagement building through the year. The PRA has also written directly to CFOs with thematic feedback on IFRS 9 ECL and the accounting for climate risk, signalling that this is now a live audit and reporting issue rather than a strategic aspiration.

The limits of post-model management overlays

Almost every bank currently reflects climate risk through a management overlay, sometimes called a post-model adjustment (PMA). This is understandable: models built on historical loss data cannot see a risk that has not yet materialised at scale, so a judgement-based top-up is a pragmatic bridge. In practice the resulting amounts have often been marginal, driven largely by transition risk in a handful of priority sectors. The concern raised by the ECB is not that overlays exist, but that many are applied at total ECL level without distinguishing the underlying drivers. An overlay that does not separate probability of default (PD) from loss given default (LGD) lacks risk sensitivity, is hard to challenge or back-test, and tends to be governed loosely. Supervisors want overlays to be a temporary, well-documented and evidenced measure with a credible path towards in-model treatment, not a permanent black box.

Practical methods for embedding physical and transition risk

A workable approach usually layers several techniques. Scenario overlays anchor the exercise in recognised pathways, most commonly the Network for Greening the Financial System (NGFS) scenarios, translating orderly, disorderly and hot-house-world narratives into macroeconomic and sectoral variables that feed ECL. Sector and geography heatmaps then prioritise effort, flagging exposures such as oil and gas, power, steel, mining, shipping and automotive for transition risk, and mortgage or commercial property books in flood- and heat-exposed locations for physical risk. The more advanced step is adjusting the risk parameters directly: uplifting PD for carbon-intensive borrowers whose business models face policy pressure, and adjusting LGD where physical hazards impair collateral values or recovery. Sensitivity analysis rounds this out, showing how ECL responds across scenarios and time horizons and helping management and auditors gauge whether provisions are directionally credible. Leading banks are already spreading PD and LGD across mortgage portfolios and elevated-risk wholesale sectors rather than relying on a single blanket number.

The data challenge behind credible climate ECL

Data remains the binding constraint. Meaningful parameter adjustments require counterparty-level emissions and transition information, asset-level location and hazard data, and forward views that stretch well beyond the typical ECL horizon. Much of this is incomplete, inconsistent or self-reported, and physical risk in particular demands geospatial data that many banks are only beginning to source. Long-dated climate effects also sit awkwardly against the twelve-month and lifetime measurement windows of IFRS 9, and against historical loss series that contain little climate signal. The pragmatic response is to build a documented data lineage, use proxies transparently where direct data is missing, and treat data quality as a governed part of the estimate rather than a caveat buried in a footnote. Strengthening governance around how overlays and assumptions are set is, in the regulators' view, as important as the modelling itself.

IFRS 7 and climate-related disclosure implications

Measurement and disclosure move together. IFRS 7 requires banks to explain the inputs, assumptions and estimation techniques behind ECL, and the significant judgements involved. Where climate risk materially affects the estimate, that includes describing the scenarios used, the sectors and geographies most exposed, the size and rationale of any overlay, and the sensitivity of ECL to key climate assumptions. Commentators, including the ICAEW, have highlighted that climate considerations are increasingly relevant to IFRS 9 and IFRS 7 disclosures even when the quantitative impact is currently small, because users need to understand how the risk is being managed. These disclosures also need to be consistent with wider climate reporting under frameworks such as IFRS S2, so that the credit provisioning story and the strategic climate narrative do not contradict each other.

For most banks the honest position today is a modest, overlay-driven climate adjustment supported by evolving models and imperfect data. That is a defensible starting point, but not a stable destination. The clear expectation from Basel, the ECB and the PRA is a steady migration from judgement-based top-ups towards risk-sensitive, PD- and LGD-level estimation, backed by robust data, strong governance and transparent disclosure. Banks that treat 2026 as the year to build that infrastructure, rather than to refine a single overlay number, will be far better placed as supervisory scrutiny intensifies.

Key takeaways

• IFRS 9 already requires forward-looking information in ECL, and supervisors now treat both physical and transition climate risk as in-scope drivers that must be reflected credibly.
• The PRA's SS5/25, finalised in December 2025, replaces SS3/19 and expects UK banks and insurers to complete a gap analysis by 3 June 2026, putting climate risk on par with other financial risks.
• The ECB's July 2024 report warns that overlays applied at total ECL level, without separating PD and LGD, lack risk sensitivity and should migrate towards in-model treatment.
• Most banks still rely on post-model management overlays, often marginal and transition-driven; these should be temporary, well-documented and evidenced rather than permanent.
• Practical methods combine NGFS scenario overlays, sector and geography heatmaps, direct PD/LGD adjustments for exposed portfolios, and sensitivity analysis across horizons.
• Data quality is the binding constraint, and IFRS 7 disclosures must transparently explain climate scenarios, assumptions, overlays and sensitivities, consistent with wider IFRS S2 reporting.