Regulators including the European Central Bank, the Bank of England, and the Basel Committee have made clear that climate risk must be reflected in banks' and lenders' expected credit loss models. The challenge for finance and risk teams is translating that principle into a concrete, auditable adjustment to IFRS 9 PD, LGD, and EAD inputs — without the benefit of clear standard-setter guidance and with limited historical data.
Why Climate Risk Belongs in ECL Models
IFRS 9 requires ECL estimates to reflect "reasonable and supportable information about past events, current conditions, and forecasts of future economic conditions." Climate risk — both physical risk (floods, droughts, extreme weather) and transition risk (carbon pricing, stranded assets, regulatory change) — affects the creditworthiness of borrowers over loan horizons. Ignoring it is increasingly difficult to defend to auditors, prudential supervisors, and investors.
The NGFS (Network for Greening the Financial System) has published a suite of climate scenarios that are now widely used as a reference framework for financial risk assessment:
- Orderly transition — early, coordinated policy action limits warming; transition costs are manageable
- Disorderly transition — delayed policy action followed by abrupt change; higher transition risk
- Hot house world — insufficient policy action; severe physical risk materialises over long horizons
Physical vs Transition Risk: Different Models Required
Physical and transition risk affect credit quality through different channels and over different time horizons, requiring separate modelling approaches:
- Physical risk — relevant for real estate lending, agriculture, infrastructure. Chronic physical risk (sea level rise, heat stress) affects long-term collateral values and borrower income. Acute physical risk (flood, wildfire) can cause sudden credit deterioration. Geospatial data and climate hazard mapping are key inputs.
- Transition risk — most relevant for carbon-intensive sectors (energy, transport, heavy industry). Carbon pricing, regulatory change, and technology disruption can impair borrower cash flows and asset values. Sector-level carbon intensity and exposure to regulatory change drive the adjustment.
"Climate risk does not require a separate ECL model — it requires a structured overlay on existing PD and LGD inputs, supported by scenario analysis and documented assumptions."
Practical Adjustment Approaches
Most institutions are currently applying climate risk as a forward-looking macroeconomic overlay rather than rebuilding PD models from scratch. Common approaches include:
- Sector-level PD stress — applying upward adjustments to PDs for high-transition-risk sectors under adverse climate scenarios, calibrated to carbon price pathways and regulatory timelines
- Collateral value haircuts — adjusting LGD for real estate portfolios in high physical-risk locations using flood or heat hazard scoring
- Scenario probability weighting — running ECL under multiple climate scenarios and weighting outcomes, analogous to the macroeconomic scenario approach already used in IFRS 9
- Management overlays — where quantitative models are not yet available, documented management overlays with explicit assumptions and sensitivity analysis
Documentation and Audit Requirements
Whatever approach is adopted, the documentation standard is high. Auditors and supervisors will expect to see:
- A clear description of how climate risk has been assessed for each material portfolio segment
- The climate scenarios used and their source (e.g., NGFS scenarios)
- The methodology for translating scenario outputs into PD, LGD, or EAD adjustments
- Sensitivity analysis showing the range of ECL outcomes across scenarios
- An explanation of why any material climate risk has been assessed as not material to the ECL estimate
Institutions that have not yet started this process should treat the initial documentation exercise as a gap analysis — identifying where data is available, where proxies are needed, and where the modelling methodology needs to be developed.
The Regulatory Direction of Travel
Supervisory expectations are moving in one direction: greater specificity, more quantification, and less reliance on narrative-only disclosures. Institutions that build climate risk into their ECL models now — even with simplified initial approaches — will be better placed to meet the next wave of supervisory requirements than those waiting for a perfect methodology that may never exist.