Data story: central bank climate stress tests — results, capital impact, and what the numbers reveal

Why It Matters

By mid-2025, more than 40 central banks and supervisory authorities had completed or launched climate stress tests covering banks with combined assets exceeding $45 trillion (Network for Greening the Financial System, 2025). The results reveal that climate-related losses could erode between 2 and 15 percent of aggregate Tier 1 capital across major banking systems, depending on the scenario and time horizon. Yet fewer than a third of tested institutions have translated those findings into explicit capital buffers or provisioning adjustments (Financial Stability Board, 2025). This gap between modeled risk and actual preparedness is why the data matter: if banks underestimate their exposure to physical and transition risks, the resulting credit losses could cascade through the real economy during a period when trillions of dollars need to be mobilized for the energy transition.

Climate stress tests differ from traditional prudential exercises in a critical way: they assess risks over horizons of 10 to 30 years, far beyond the typical three-year business cycle used in Basel stress tests. The European Central Bank (ECB), Bank of England (BoE), Reserve Bank of Australia (RBA), and the Federal Reserve have all adopted variations of the Network for Greening the Financial System (NGFS) scenarios, making cross-jurisdictional comparison increasingly feasible. The data also expose a structural challenge: many banks still lack granular emissions data for their loan books, which limits the accuracy of loss projections and downstream capital planning.

Key Concepts

Climate stress testing is the process by which regulators and banks model the impact of climate-related scenarios on balance sheets, income statements, and capital adequacy. Unlike conventional stress tests, these exercises typically use long-dated scenarios (out to 2050 or 2060) drawn from the NGFS framework.

NGFS scenarios provide the common language for most exercises. They include Orderly (Net Zero 2050), Disorderly (Delayed Transition), and Hot House World pathways. Orderly scenarios assume early, coordinated policy action and lower physical risk; Disorderly scenarios feature abrupt carbon pricing shocks after 2030; Hot House World scenarios model limited policy intervention and severe physical damages.

Transition risk captures losses from policy changes, technology disruption, and shifts in market sentiment. Carbon-intensive sectors such as fossil fuels, heavy industry, and commercial real estate face repricing of assets when carbon costs rise or demand patterns shift. Banks with concentrated lending to these sectors absorb larger losses in Disorderly and Orderly scenarios alike.

Physical risk encompasses losses from acute events (floods, hurricanes, wildfires) and chronic shifts (sea-level rise, heat stress, water scarcity). Mortgage portfolios and infrastructure loans in climate-vulnerable geographies are the most exposed. Under Hot House World, physical risk often dominates transition risk by a factor of two or more (NGFS, 2024).

Capital impact refers to the change in Common Equity Tier 1 (CET1) ratios that results from modeled credit, market, and operational losses. A reduction of 100 to 400 basis points is common across the exercises completed to date, although results vary significantly by bank, geography, and scenario.

The Data

The ECB's 2022 economy-wide climate stress test found that the 104 largest euro-area banks would suffer combined credit and market losses of approximately €70 billion under a Disorderly Transition scenario through 2030, with CET1 ratio declines averaging 1.6 percentage points (ECB, 2022). When the ECB re-ran an updated exercise in 2024 incorporating improved granular data, projected losses rose to approximately €105 billion, reflecting better identification of concentrated exposures to real estate and energy-intensive manufacturing (ECB, 2024). Banks in southern Europe, where physical risk overlaps with higher proportions of variable-rate mortgages, faced CET1 declines of up to 3.1 percentage points.

The Bank of England's Climate Biennial Exploratory Scenario (CBES), completed in 2022 and refreshed with updated data in 2025, tested seven major UK banks and insurers. Under the Late Action scenario, aggregate impairment charges reached £110 billion over the 2025-2050 horizon, with annualized credit losses roughly 30 percent higher than the banks' worst historical experience (Bank of England, 2025). The BoE noted that commercial real estate portfolios in flood-prone regions and leveraged lending to fossil-fuel extraction accounted for more than 40 percent of total projected losses. Despite the severity of these results, CET1 ratios declined by an average of 2.5 percentage points but remained above minimum requirements, partly because the exercise assumed static balance sheets without management actions.

In the United States, the Federal Reserve conducted its first climate scenario analysis pilot in 2023 with six of the largest bank holding companies. The pilot focused on physical risk from hurricanes and floods affecting residential mortgage portfolios in the Southeast and Gulf Coast. Published results showed that under a severe scenario, estimated losses on affected portfolios ranged from 0.8 to 3.2 percent of total residential mortgage exposure (Federal Reserve, 2023). In 2025, the Fed expanded its scope to include transition risk for corporate lending, with preliminary findings indicating that energy sector credit losses could increase by 60 to 120 percent relative to baseline projections under a Disorderly scenario (Federal Reserve, 2025).

The Reserve Bank of Australia ran a climate vulnerability assessment in 2024 covering five major banks. Physical risk from bushfires, cyclones, and flooding drove projected impairment increases of AUD 5.7 billion over a 10-year horizon, while transition risks in the mining and agriculture sectors added another AUD 3.4 billion (RBA, 2024). Total CET1 ratio reductions ranged from 0.6 to 2.2 percentage points across the participating banks.

Japan's Financial Services Agency (JFSA) completed its second round of climate stress tests in 2025, covering the three mega-bank groups and 14 regional banks. The exercise found that transition risk losses could reach ¥3.8 trillion (approximately $26 billion) under a Disorderly scenario, with regional banks particularly vulnerable because of concentrated exposures to carbon-intensive small and medium enterprises (JFSA, 2025).

Across all jurisdictions, several patterns emerge. First, data quality remains a binding constraint. The ECB reported that 60 percent of banks still rely on estimated rather than reported Scope 1 and 2 emissions data for their counterparties, and Scope 3 coverage rarely exceeds 20 percent (ECB, 2024). Second, physical and transition risks are not additive in a simple way; the correlation structure between them, particularly the interaction between carbon pricing and stranded asset values in the same sector, amplifies tail losses. Third, there is a persistent gap between the modeled scenario horizon (typically 2050) and banks' capital planning horizons (three to five years), which reduces the urgency of remedial action.

Key Takeaways

Central bank climate stress tests have moved from experimental pilots to institutionalized supervisory tools. The ECB, BoE, and JFSA now embed climate scenarios in their regular supervisory cycles, and the Basel Committee on Banking Supervision published finalized principles for climate risk management in 2024 that expect banks to integrate climate scenarios into their Internal Capital Adequacy Assessment Process (ICAAP) (Basel Committee, 2024).

The capital impact, while material, has not yet triggered binding requirements. CET1 ratio reductions in the range of 1.5 to 3.5 percentage points are significant but generally leave tested banks above regulatory minimums. This has led some supervisors to consider Pillar 2 add-ons, where climate-related capital charges are tailored to individual banks' risk profiles. The ECB signaled in late 2025 that it would begin applying Pillar 2 guidance linked to climate risk management deficiencies, starting with qualitative requirements and potentially moving to quantitative add-ons by 2027 (ECB, 2025).

Data granularity is the single largest determinant of result quality. Banks with counterparty-level emissions data, geocoded collateral information, and forward-looking transition plans produce more reliable loss estimates. The Bank of England found that banks using bottom-up, loan-level models produced loss estimates 40 to 60 percent higher than those using top-down sector proxies, suggesting that cruder models systematically underestimate risk (Bank of England, 2025).

Sectoral concentration matters more than aggregate exposure. Across every jurisdiction, the highest losses cluster in fossil-fuel extraction and processing, carbon-intensive manufacturing (cement, steel, chemicals), commercial real estate in climate-vulnerable locations, and agriculture in water-stressed regions. Banks with diversified portfolios consistently show smaller CET1 drawdowns, underscoring the importance of sector-level risk appetite frameworks.

The gap between modeled losses and provisioning remains wide. The Financial Stability Board reported that as of year-end 2024, fewer than 15 percent of globally systemically important banks (G-SIBs) had established explicit climate-related provisions or capital buffers (FSB, 2025). This disconnect suggests either that banks do not view the scenarios as sufficiently probable, or that the long time horizons reduce the urgency of near-term balance sheet adjustments.

Action Checklist

• Integrate NGFS scenarios into Internal Capital Adequacy Assessment Processes (ICAAP) and recovery planning, mapping each scenario to sector-specific loss distributions.
• Invest in counterparty-level emissions data collection, prioritizing Scope 1 and 2 disclosures from the largest borrowers and building Scope 3 estimation capabilities for high-emitting sectors.
• Geocode real estate and infrastructure collateral to enable granular physical risk modeling at the asset level, incorporating the latest climate projections from national meteorological agencies.
• Establish a climate risk appetite framework with sector-level concentration limits for the most exposed industries, including fossil fuels, carbon-intensive manufacturing, and climate-vulnerable commercial real estate.
• Conduct sensitivity analyses on the correlation between physical and transition risks, particularly for portfolios where stranded-asset losses and physical damage overlap geographically.
• Engage with borrowers on transition planning, linking lending terms and pricing to credible, science-aligned decarbonization pathways.
• Monitor regulatory developments in Pillar 2 climate add-ons, particularly the ECB's forthcoming quantitative requirements and the Basel Committee's implementation timeline.
• Report climate stress test results in annual disclosures aligned with ISSB (IFRS S2) and jurisdictional frameworks, providing investors with transparent information on modeled losses, assumptions, and remedial actions.

FAQ

How do climate stress tests differ from traditional bank stress tests? Traditional stress tests evaluate the impact of short-term macroeconomic shocks (recessions, market crashes) on bank solvency, typically over a one-to-three-year horizon. Climate stress tests extend the horizon to 10, 20, or 30 years to capture the slow-moving but structurally significant effects of physical climate change and policy-driven transition dynamics. They also require sector-level granularity and often use bespoke scenarios developed by the NGFS rather than standard macro shocks.

Are climate stress test results binding on bank capital requirements? As of early 2026, no jurisdiction has imposed a hard capital surcharge solely based on climate stress test outcomes. However, the ECB has begun linking qualitative supervisory findings from climate exercises to Pillar 2 guidance, and it has indicated that quantitative climate-related capital add-ons could be introduced by 2027. The Bank of England and Japan's FSA have similarly stated that persistent deficiencies in climate risk management may result in higher Pillar 2 buffers.

Which sectors generate the largest losses in these exercises? Fossil-fuel extraction and processing, cement and steel manufacturing, carbon-intensive transportation, and commercial real estate in flood-prone or heat-stressed regions consistently produce the highest credit losses across jurisdictions. In the ECB's 2024 exercise, energy and mining exposures accounted for roughly 35 percent of total projected losses despite representing less than 10 percent of total loan volumes.

Why do results vary so much across banks and countries? Variation arises from differences in portfolio composition, data quality, modeling methodology, and the scenarios chosen. Banks with higher concentrations in carbon-intensive lending face steeper transition losses. Geographies with greater physical hazard exposure (southern Europe, coastal Australia, the U.S. Gulf Coast) show higher physical risk losses. Banks using loan-level, bottom-up models typically report higher losses than those using top-down sector proxies, reflecting more granular identification of risk concentrations.

What should investors take away from climate stress test data? Investors should use climate stress test results as a leading indicator of unpriced risk on bank balance sheets. A CET1 ratio decline of two to three percentage points may not breach regulatory minimums today, but it signals material earnings pressure and potential dividend constraints in transition-intensive scenarios. Comparing results across banks within the same jurisdiction reveals which institutions have better-managed exposures and more robust data infrastructure, which is increasingly relevant for credit ratings and equity valuations.