Climate scenario analysis: from strategic exercise to binding requirement

In force from 3 December 2025, the Prudential Regulation Authority (PRA) Supervisory Statement SS5/25 has raised the bar for climate scenario analysis for UK banks and insurers. Climate scenarios must now integrate directly into IFRS 9 expected credit loss (ECL) modelling and Internal Capital Adequacy Assessment Process (ICAAP) submissions, a process our ICAAP guide sets out in full.¹ Firms had six months from commencement, to 3 June 2026, to review their own position against the new expectations and to hold a plan for closing any gaps they found.²

How is climate scenario analysis different from a stress test?

Climate scenario analysis differs from conventional macroeconomic stress testing in three fundamental ways, and each difference has methodological implications.

1. Time horizon. Traditional stress tests operate over one to three years, where historical relationships between macroeconomic variables and default rates exist. Climate scenarios extend to 2050 and beyond (the Bank of England’s 2021 climate exercise asked banks to project losses over 30 years³), covering transitions and physical impacts that might have no relationship in historical credit cycles.
2. Nature of uncertainty. Standard stress tests are calibrated against historical experience. Climate outcomes are harder to pin down: how fast the energy transition will happen, how much physical damage global warming will cause, or how governments will respond. The spread is wide enough that NGFS does not try to close it. Its orderly narratives are pinned to end-of-century targets of 1.5°C and 2°C, while its hot house narratives carry no temperature target at all, only high physical risk.⁴ None of that can be modelled as neatly as GDP growth.
3. Directionality. In a conventional stress test, an adverse shock is usually followed by recovery. Climate risk operates differently. Under a high-emissions scenario, physical damage accumulates permanently, and economic indicators don’t necessarily bounce back quickly, or at all.

The differences in time horizon, uncertainty and directionality mean firms cannot simply stretch their existing stress-testing scenarios to cover climate risk. They need scenarios built for the purpose.

The Network for Greening the Financial System (NGFS) has developed a global reference framework for climate scenario analysis. It is a coalition of 152 central banks and supervisors as at April 2026.⁵ Its Phase V scenarios, published in November 2024, map seven narratives across four outcomes. Each scenario translates climate policy assumptions into the macroeconomic variables (GDP growth, inflation, interest rates, equity prices) that feed directly into conventional credit models.

SS5/25 names no framework. It lets firms use scenarios that are “either externally or internally developed according to the materiality of the risk and the level of internal expertise”, and asks instead that the models be conceptually sound, supported by published scientific, technological and economic research, and that firms be able to justify the sources they relied on.² NGFS is simply where most firms go, and the PRA’s own footnotes point to its guidance on where institutional adaptation is needed.⁶

What do the Basel climate principles require of UK banks under SS5/25?

Two instruments set the expectations UK banks work to: the Basel Committee’s climate principles internationally, and the PRA’s SS5/25 domestically. NGFS supplies the scenarios; these two say what a firm must do with them.

The international standard is the Basel Committee on Banking Supervision (BCBS): Principles for the effective management and supervision of climate-related financial risks, which sets 18 high-level principles: 12 directed at banks, six at supervisors. Two bank-facing principles are directly relevant:

• Principle 12 addresses banks, not supervisors: “Where appropriate, banks should make use of scenario analysis to assess the resilience of their business models and strategies to a range of plausible climate-related pathways.”⁷ The supervisor-facing counterpart is Principle 15, which asks supervisors to determine that banks apply it.
• Principle 5 requires firms to incorporate climate risk into their capital and liquidity adequacy frameworks, including the ICAAP and stress-testing programmes.

The Basel Committee’s bank-facing and supervisor-facing principles both apply proportionately through national supervisory regimes.

In the UK, the operative instrument is now the PRA’s SS5/25, which superseded SS3/19 on 3 December 2025, and our piece on the PRA raising the bar on climate risk works through what it asks for.⁸ Where SS3/19 required scenario analysis as a strategic planning tool, SS5/25 demands that climate scenario analysis feeds into IFRS 9 ECL provisioning and the ICAAP.

“Firms should identify the expectations that require further work for them to meet, and develop a plan for how they will address, any gaps.”PRA SS5/25, paragraph 3.2²

The stress-test evidence: why embedding can’t wait

SS5/25’s demands follow directly from what supervisors saw when they tested the system. Supervisory exercises in the EU and the UK agree on two things: climate-related losses are material across every scenario tested, and severity scales with how abrupt or delayed the transition path turns out to be. The UK exercise makes the timing point most directly, because its scenarios differ by exactly that.

Under SS5/25, the CBES and EU findings are no longer background reading: a loss distribution this sensitive to transition timing is exactly what supervisors now expect to see reflected in ECL provisions and ICAAP capital assessments.

When climate risk outruns the ECL model

Integrating climate scenarios into IFRS 9 ECL models is hardest where their time horizons diverge.

IFRS 9 requires that forward-looking assumptions be reasonable and supportable, and available without undue cost or effort. IFRS 9 sets no fixed horizon.¹⁰ What it sets is a limit on credibility: beyond the period a firm’s detailed macroeconomic forecasts cover, the standard expects extrapolation rather than projection. Climate transition risk mostly shows up after that point. The Bank of England’s CBES losses that clustered following the 2031 delay fall well outside anything an ECL forecast reaches directly. It isn’t realistic to expect ECL models to project macro conditions across a 25-year mortgage book with the same precision as a three-year model.

Most firms currently bridge that gap through post-model adjustments (PMAs), expert-judgement adjustments applied on top of model-output ECL where the core model does not yet capture a known risk driver.

PMAs are an interim solution, not a permanent one, and the European Banking Authority (EBA) guidelines on accounting for expected credit losses say so directly: institutions should use them “only as an interim solution”, and where the risk driver is not transient “the methodology should be updated in the near term to incorporate the factor”.¹¹ The ECB has since set out how supervisors expect overlays for risks outside the model to be built and governed.¹² SS5/25 brings the same discipline to climate risk specifically.

What regulators require now

Having a climate scenario capable of supporting a board narrative is no longer sufficient. With the EU exercise putting losses under a combined climate stress at nearly double its own baseline, the regulation now requires a set of minimum auditable properties:

• Scenario choices must be justifiable: conceptually sound models, published research behind them, and a documented rationale for the sources relied on, whether the scenarios came from NGFS or were built in-house.
• Climate scenario outcomes must be translated to credit risk parameters.
• PMAs must be temporary: where the risk driver is not transient, the model itself has to be updated in the near term.
• Materiality criteria must be clearly defined and documented, not just applied.² EIOPA’s application guidance sets out the same discipline for insurers running the assessment through the ORSA.¹³

Climate scenario analysis is entering the same review cycle that stress testing entered a decade ago, from conceptual exercise to operational discipline.

For a firm still leaning on post-model adjustments, the binding question is no longer whether to build the capability, but whether the remediation plan on file would survive an SS5/25 review.

Frequently asked questions