Deep dive: Climate risk stress testing & scenario regulation — what's working, what's not, and what's next

The European Central Bank's 2024 climate stress test revealed that eurozone banks face potential losses of EUR 70 billion under a disorderly transition scenario, yet fewer than 30% of tested institutions had integrated climate variables into their core risk management frameworks (ECB, 2024). That gap between regulatory ambition and institutional readiness defines the current state of climate risk stress testing globally. Central banks in 46 jurisdictions have now completed or announced climate stress testing exercises, up from just 8 in 2020, with aggregate assets under supervision exceeding $140 trillion (Network for Greening the Financial System, 2025). For sustainability professionals navigating this rapidly evolving regulatory landscape, understanding which approaches are delivering actionable results and which are producing compliance theater is critical for strategic planning.

Why It Matters

Climate-related financial risks threaten the stability of the global financial system across two primary channels: physical risks from extreme weather events and chronic climate shifts, and transition risks from policy changes, technology disruptions, and shifting market preferences. The Bank for International Settlements estimates that unmanaged climate risks could trigger asset repricing of $4 trillion to $20 trillion depending on the transition pathway, with concentrated exposures in fossil fuel assets, coastal real estate, and carbon-intensive manufacturing (BIS, 2025).

Regulatory momentum has accelerated dramatically. The Network for Greening the Financial System (NGFS) now counts 134 members representing central banks and supervisors covering 88% of global GDP. The European Banking Authority requires all EU-supervised banks to incorporate climate scenarios into their Internal Capital Adequacy Assessment Process (ICAAP) by 2026. The Bank of England's Climate Biennial Exploratory Scenario (CBES) has completed two full cycles, and the Federal Reserve conducted its first pilot climate scenario analysis in 2023 with expanded exercises planned through 2027.

The financial materiality is becoming undeniable. Swiss Re estimates that climate-related insured losses exceeded $130 billion in 2025 alone, and the protection gap (uninsured losses) reached $280 billion (Swiss Re, 2026). Financial institutions that fail to accurately model and price these risks face both balance sheet surprises and regulatory penalties as supervisory expectations harden from guidance into binding requirements.

Key Concepts

Climate scenario analysis involves constructing plausible future states of the world that combine climate science projections (temperature pathways, physical hazard frequencies) with economic and policy assumptions (carbon pricing trajectories, technology adoption curves, stranded asset timelines). The NGFS provides six reference scenarios ranging from orderly transition (Net Zero 2050) to hot house world (current policies), each specifying hundreds of macroeconomic and sectoral variables across 50-year time horizons. Institutions translate these scenarios into impacts on their specific portfolios through exposure mapping, vulnerability assessment, and loss estimation.

Climate Value-at-Risk (CVaR) extends traditional VaR methodologies to capture climate-specific risk factors, estimating the maximum expected loss attributable to climate-related events over a defined time horizon and confidence interval. Unlike traditional VaR, CVaR must account for fat-tailed physical risk distributions, path-dependent transition dynamics, and correlations between climate hazards that differ fundamentally from historical financial market correlations. Leading implementations combine bottom-up asset-level analysis with top-down macroeconomic scenarios.

Transition risk modeling quantifies the financial impact of policy, technology, and market shifts associated with the low-carbon transition. Key variables include carbon price trajectories ($50 to $250 per tonne by 2030 depending on jurisdiction and scenario), stranded asset timelines for fossil fuel reserves and infrastructure, and demand destruction curves for carbon-intensive products. The challenge lies in capturing nonlinear tipping points: a carbon price increase from $80 to $120 per tonne may have minimal impact on most sectors but render certain coal-dependent utilities immediately insolvent.

Physical risk assessment maps financial exposures to location-specific climate hazards including flooding, wildfire, hurricane wind speed, heat stress, water scarcity, and sea level rise. Granular physical risk models operate at the individual asset level, combining geocoded exposure data with climate hazard projections at resolutions of 1 to 5 km. The temporal dimension is critical: a 30-year mortgage originated in 2026 in a coastal Florida community faces meaningfully different physical risk in years 20 to 30 than in years 1 to 5.

What's Working

Central Bank Supervisory Exercises

The Bank of England's Climate Biennial Exploratory Scenario (CBES) has emerged as the gold standard for supervisory climate stress testing. The second CBES cycle in 2024 covered 19 major banks and insurers representing over 80% of UK financial sector assets. Participating institutions reported that the exercise drove measurable improvements in data collection, governance structures, and internal capabilities. Barclays disclosed that its CBES participation led to the creation of a dedicated climate risk analytics team of 45 professionals and the integration of climate variables into 12 of its 17 material risk categories. HSBC reported developing sector-specific transition risk models for its 10 most carbon-intensive lending sectors, enabling portfolio-level carbon price sensitivity analysis for the first time.

The ECB's climate stress test covered 104 significant institutions across the eurozone and produced the most granular publicly available dataset on European bank climate exposures. The exercise revealed that banks with higher exposure to carbon-intensive sectors faced projected credit losses 1.5 to 2.3 times greater under the disorderly transition scenario compared to the orderly pathway. The ECB used results to issue institution-specific supervisory expectations and set deadlines for remediation, creating direct links between stress test performance and supervisory outcomes.

Sector-Specific Transition Risk Tools

The Paris Agreement Capital Transition Assessment (PACTA) tool, developed by the Rocky Mountain Institute, has been adopted by over 4,600 financial institutions globally to assess portfolio alignment with climate scenarios. PACTA covers 8 key climate-relevant sectors (power, oil and gas, coal, automotive, cement, steel, aviation, and shipping) and maps institutional exposures to asset-level production data. The Swiss Federal Office for the Environment used PACTA to conduct a nationwide assessment of the Swiss financial sector's climate alignment, covering 79% of domestically managed assets and revealing that Swiss bank lending portfolios were on track for a 2.7 degrees C warming pathway rather than the stated 1.5 degrees C commitments.

The Transition Pathway Initiative (TPI), managed by the Grantham Research Institute at the London School of Economics, provides carbon performance benchmarks for 600 companies across the highest-emitting sectors. Asset managers managing over $60 trillion in assets use TPI benchmarks to evaluate investee company alignment with Paris Agreement goals. The tool's management quality assessment scores, which evaluate governance, strategy, risk management, and target-setting quality, have demonstrated predictive power: companies scoring in the top quartile on TPI management quality delivered 12% higher risk-adjusted returns over 5-year periods compared to bottom-quartile peers (TPI, 2025).

Open-Source Physical Risk Platforms

OS-Climate, a Linux Foundation initiative backed by Allianz, Amazon, BNP Paribas, and Goldman Sachs, has developed open-source physical risk analytics covering hazard, vulnerability, and financial impact assessment. The platform provides asset-level physical risk scores for over 50 million geocoded assets globally, combining 14 climate hazard variables with structural vulnerability functions and financial loss models. BNP Paribas reported that integrating OS-Climate physical risk scores into its real estate lending portfolio identified $2.3 billion in exposures requiring enhanced monitoring, with 340 individual properties flagged for climate-adjusted loan-to-value reassessment.

What's Not Working

Data Gaps and Quality Issues

Climate stress testing is fundamentally constrained by data availability and quality. Counterparty-level emissions data remains incomplete: fewer than 40% of public companies and fewer than 5% of private companies disclose Scope 1, 2, and 3 emissions with third-party verification (CDP, 2025). Financial institutions relying on estimated emissions data face uncertainty ranges of 30 to 60% for Scope 1 and 2, and 50 to 200% for Scope 3. This uncertainty propagates through stress test models, producing loss estimates with confidence intervals so wide that they offer limited decision-making value. The Bank of England's CBES review acknowledged that participating institutions flagged data quality as their number one challenge, with some banks reporting that they could only map 55 to 70% of their corporate loan books to emissions-intensive sector classifications.

Geospatial data for physical risk assessment also has material gaps. Flood risk models in developing economies rely on limited historical observations and coarse-resolution terrain data, producing hazard estimates with error margins of 2 to 5 times. The Insurance Development Forum estimates that 97% of insured assets in Sub-Saharan Africa and 85% in South and Southeast Asia lack granular physical risk assessments.

Methodological Inconsistencies

The absence of standardized stress testing methodologies produces results that are difficult to compare across jurisdictions and institutions. The ECB, Bank of England, Federal Reserve, and Bank of Japan each use different scenario specifications, time horizons (ranging from 3 years to 50 years), and loss estimation approaches. A 2025 comparison by the Financial Stability Board found that the same portfolio of European bank assets produced estimated climate-related losses ranging from 1.2% to 8.7% of risk-weighted assets depending on the methodology applied, a sevenfold variation that undermines the credibility and comparability of results.

Within institutions, the disconnect between climate scenario analysis and traditional risk management frameworks remains acute. Climate stress tests typically run on separate models, use different time horizons, and produce outputs that do not integrate into existing capital planning processes. The result is that climate stress test findings often remain siloed in sustainability or risk research teams rather than influencing actual capital allocation, pricing, or provisioning decisions.

Short Time Horizons

Regulatory stress tests typically operate on 1 to 5 year horizons aligned with capital planning cycles, but the most material climate risks manifest over 10 to 30 year periods. This temporal mismatch means that stress tests systematically understate climate risk by truncating the assessment window before the most severe physical and transition impacts materialize. The NGFS has acknowledged this limitation but has not yet proposed a framework for integrating longer-term climate projections into near-term supervisory requirements. Some central banks have experimented with 30-year exploratory scenarios, but these remain disconnected from binding capital requirements.

Key Players

Established Companies

• MSCI: provides climate risk analytics covering physical and transition risk scoring for 10,000 companies and 300,000 individual assets, used by institutions managing over $30 trillion in assets
• S&P Global Sustainable1: offers climate scenario analysis tools integrating physical risk, transition risk, and carbon pricing impacts into portfolio-level financial projections
• Moody's Analytics: delivers climate risk assessment solutions combining proprietary catastrophe models with NGFS scenario-aligned transition risk analytics for banking and insurance portfolios
• Bloomberg: operates the Bloomberg Green terminal module providing climate risk data, TCFD-aligned analytics, and scenario analysis tools integrated into existing financial workflows

Startups

• Jupiter Intelligence: specializes in hyperlocal physical climate risk analytics using AI-enhanced climate models at 90-meter resolution for asset-level risk scoring
• Risilience: provides enterprise climate scenario planning software that integrates transition and physical risk into financial planning and strategy formulation
• Cervest: offers EarthScan, an AI-driven platform providing asset-level physical climate risk intelligence covering 6 hazard categories with forward-looking projections to 2100

Investors

• Glasgow Financial Alliance for Net Zero (GFANZ): coordinates climate risk assessment practices across financial institutions managing over $130 trillion in assets
• Network for Greening the Financial System (NGFS): consortium of 134 central banks driving supervisory climate stress testing standards and scenario development
• Climate Finance Leadership Initiative: convened by Michael Bloomberg, mobilizing private capital for climate risk analytics and stress testing infrastructure

KPI Benchmarks by Use Case

Metric	Banking	Insurance	Asset Management
Portfolio coverage (% assessed)	60-85%	70-90%	50-75%
Scenario count (minimum)	3-6	4-8	3-5
Time horizon (years)	3-30	5-50	5-30
Data quality score (% verified)	35-55%	40-60%	30-50%
Physical risk assets mapped	50-75%	70-95%	40-65%
Update frequency	Annual	Semi-annual	Quarterly
Integration into capital planning	20-40%	30-50%	15-35%

Action Checklist

• Map existing portfolio exposures to NGFS climate-relevant sectors and establish baseline emissions intensity metrics for each material sector
• Select and implement at least three NGFS reference scenarios (orderly, disorderly, and hot house world) as the foundation for internal climate scenario analysis
• Assess current data coverage and quality gaps for counterparty-level emissions, geocoded asset locations, and sector classification accuracy
• Integrate physical risk scoring into real estate and infrastructure lending or investment processes using asset-level hazard data at sub-5 km resolution
• Establish governance structures linking climate stress test outputs to capital allocation, pricing, and provisioning decisions rather than treating them as standalone compliance exercises
• Build internal climate risk analytics capability with dedicated staff trained in both climate science and financial risk modeling
• Engage with regulators proactively to understand upcoming supervisory expectations and timeline for mandatory climate stress testing requirements
• Develop transition risk models for top 5 to 10 carbon-intensive sector exposures with carbon price sensitivity analysis at $50, $100, $150, and $200 per tonne

FAQ

Q: What is the difference between climate stress testing and traditional financial stress testing? A: Traditional stress tests evaluate portfolio resilience to macroeconomic shocks (recession, interest rate spikes, market crashes) over 1 to 3 year horizons using historical data and statistical models. Climate stress tests differ in three fundamental ways: they require forward-looking scenarios derived from climate science rather than historical patterns, they operate over much longer time horizons (10 to 50 years) to capture the full materialization of physical and transition risks, and they must account for unprecedented nonlinear dynamics including tipping points, feedback loops, and systemic cascading failures that have no historical precedent in financial data.

Q: Are climate stress test results currently used for regulatory capital requirements? A: Not directly in most jurisdictions as of early 2026, but the trajectory is toward integration. The ECB has stated that climate stress test results inform the Supervisory Review and Evaluation Process (SREP) and can lead to institution-specific Pillar 2 capital add-ons. The Bank of England uses CBES results to set supervisory expectations but has not yet linked them to binding capital requirements. The European Banking Authority's roadmap calls for incorporating climate risk into Pillar 1 capital frameworks by 2028. Financial institutions should prepare for a progression from exploratory exercises to binding capital implications over the next 2 to 4 years.

Q: How should smaller financial institutions approach climate stress testing with limited resources? A: Start with proportionate approaches. Use freely available tools such as PACTA for portfolio alignment assessment and NGFS scenarios with simplified sensitivity analysis. Focus initial efforts on material exposures: identify the 20% of the portfolio representing the highest climate risk concentration and prioritize data collection and analysis for those exposures. Leverage industry initiatives and peer benchmarks rather than building proprietary models. The NGFS provides guidance specifically for smaller institutions, recommending a phased approach that begins with qualitative scenario narratives and progresses to quantitative modeling as capabilities develop.

Q: What are the most common pitfalls in climate scenario analysis? A: Four pitfalls dominate. First, treating scenarios as forecasts rather than plausible future states, leading to false precision and overconfidence in point estimates. Second, using overly aggregated sectoral analysis rather than asset-level or counterparty-level assessment, which masks concentration risks. Third, ignoring second-order effects and cross-sector contagion: a disorderly transition does not only impact fossil fuel companies but cascades through dependent supply chains, regional economies, and labor markets. Fourth, failing to connect scenario analysis outputs to actual business decisions, resulting in expensive compliance exercises that do not change capital allocation, pricing, or strategy.

Sources

• European Central Bank. (2024). 2024 Climate Risk Stress Test: Results and Supervisory Assessment. Frankfurt: ECB Banking Supervision.
• Network for Greening the Financial System. (2025). NGFS Climate Scenarios for Central Banks and Supervisors: Technical Documentation. Paris: NGFS.
• Bank for International Settlements. (2025). Climate-Related Financial Risks: Measurement Methodologies and Supervisory Implications. Basel: BIS.
• Swiss Re Institute. (2026). Sigma: Natural Catastrophes and Climate Risk in 2025. Zurich: Swiss Re.
• CDP. (2025). Global Climate Disclosure Report 2025: Progress and Gaps in Corporate Climate Reporting. London: CDP.
• Financial Stability Board. (2025). Supervisory and Regulatory Approaches to Climate-Related Risks: Cross-Jurisdictional Comparison. Basel: FSB.
• Transition Pathway Initiative. (2025). TPI State of Transition Report 2025: Carbon Performance and Management Quality Assessment. London: Grantham Research Institute, LSE.