Climate scenario analysis for small banks: A practical guide to proportionate ESG risk management
Small banks face a fundamental challenge in climate risk management: regulatory expectations designed for large institutions with extensive resources are being applied across the entire banking sector, regardless of size or capacity. To counter this, Johannes Voit proposes a translation framework, rather than costly replications.
A small bank should not try to generate its own macroeconomic scenarios. It should construct a bank-specific scenario by translating a small number of externally generated climate drivers into portfolio-relevant economic impacts through a transparent, deliberately simplified transmission framework.
That can be both credible and proportionate.
The regulatory direction actually supports this architecture. The EBA's final environmental scenario-analysis guidelines distinguish short-term stress testing from medium-/long-term resilience analysis, while the ESG-risk guidelines emphasise proportionality and materiality. The Network for Greening the Financial System (NGFS) itself explicitly says that its scenarios are a common reference framework and may need to be adapted to the user's particular needs.
Model what is relevant, not the whole economy
A small bank can use a four-layer scenario architecture:
Theory | Example: NGFS orderly transition |
External scenario | Higher carbon price, tighter regulation, energy-price restructuring |
Relevant risk drivers | Higher costs for energy-intensive borrowers |
Transmission channels | Weaker EBITDA, investment burden, collateral effects |
Bank-specific financial impacts | Higher PD and possibly LGD |
The bank therefore doesn't need to calculate the entire chain with a structural macroeconomic model. Instead, it needs to make the last part of the chain explicit, documented and defensible. That is a much more manageable problem.
Moreover, the three horizons should use three different methodologies
Horizon | Approx. period | Main question | Appropriate methodology |
Short term | ~1 year | What could hit our balance sheet quickly? | Stress test |
Medium term | ~3–5 years | How could transition / physical risks alter credit quality? | Narrative + driver-based scenario |
Long term | ~10+ years | Is our business model resilient under alternative futures? | Narrative + structural scenario |
Trying to force a 2050 scenario into a one-year PD model is conceptually wrong. Likewise, treating a one-year climate shock merely as a qualitative "resilience assessment" wastes the information that can actually be quantified.
Ten steps to ensure a pragmatic implementation of a credible and proportionate scenario analysis
- Materiality: Identify the 5–10 portfolio segments where climate/nature risk can plausibly become financially material.
- External anchors: Select a small number of credible external scenarios from NGFS and other authoritative sources.
- Scenario compression: Reduce them to 10–15 bank-relevant macro, transition, physical and nature drivers.
- Portfolio segmentation: Map exposures into geographical and economic segments.
- Transmission channels: For each segment identify: relevant driver → dependency / vulnerability → economic impact → financial-risk channel.
- Impact functions: Use simple elasticities, thresholds and expert-validated coefficients rather than developing a macro-financial model.
- Financial translation: Feed the resulting borrower / portfolio effects into existing systems wherever quantitatively defensible: rating → PD → LGD → expected losses and capital
- Three horizons:
- 1 year: quantified stress test combining external short-term stress scenarios and bank-specific vulnerabilities
- 3–5 years: quantified / semi-quantified transition and physical scenario analysis
- 10+ years: strategic resilience and business-model analysis.
- Uncertainty: Give ranges rather than false precision.
- Governance: Document source, transformation, expert judgment, uncertainty and validation for every important assumption.
When time and resources are limited, banks should first set up a master plan specifying what is to be achieved initially, and the extensions planned in later stages. For example, in a first step, it might use only one scenario (the “central scenario”) and only the three most relevant business segments, and restrict the analysis to climate scenarios.
The importance of sharing
Sharing and co-creation are most important with emerging risks where data are scarce, and there is a huge methodological and implementation uncertainty. Banks can collaborate on topics such as data, methodology, scenario ranking (e.g., among the seven NGFS long-term scenarios), validation of assumptions and parameters (e.g. impact functions, step 6 above). The most successful example where even competing banks share and collaborate is ORX, the Operational Riskdata eXchange Association.
Often, working groups form under the auspices of bankers’ associations where much more support for small banks can be provided. For example, the German Savings Banks Association (DSGV) set up a complete set of (in some instances sophisticated) ESG risk management tools in two projects involving prominently small savings banks (alongside with interested bigger banks). It includes climate stress tests and resilience analysis along with an alignment tool for the medium-term horizon. These tools are delivered initially as Excel prototypes and integrated into the core risk management architecture later.
Influence on loan decisions
In the marketplace, we often find oversimplified heuristics when it comes to credit decisions involving ESG factors: banks fix thresholds, e.g., on energy performance certificates or ESG scores, and reject loan applications above the thresholds and approve those below the threshold at standard conditions. Obviously, this practice often is counterproductive.
Can the scenarios help in ESG-risk-adjusted pricing of loans? Basically, the central scenario is the new baseline for the future. When the medium-term analysis provides estimates for future PD and LGD variations, a pricing add-on can be constructed for the various segments. With the engagement typical of small banks, a client-specific modification of the portfolio add-on is feasible.
Set up a simple heuristic: classify clients into categories ranging from non-transformable (no idea about ESG, no interest to question business model) to winners of the transformation (knows about influence of ESG on business and has a clear strategy on how to emerge stronger), and increase/decrease the portfolio-add-on accordingly. This is in line with studies on the loan conditioning behaviour of French banks after a climate stress test performed by a national supervisor.
The central insight
Small banks do not need small versions of the NGFS models. They need a disciplined translation layer between NGFS scenarios and their own portfolios.
The external institutions provide the physics, transition pathways and macroeconomic context. The bank provides the portfolio exposure, vulnerabilities, transmission channels and financial consequences.
That division of labour is both economically sensible and potentially much more scalable than expecting every smaller bank to build a climate-econometric infrastructure.
Discover climate risk management best practices with professionals across banking, asset and investment management, regulatory bodies and academia at RiskMinds this November!
Disclosure
For several months, the author has had almost daily conversations with ChatGPT on scenario analysis, mostly in an effort to deconstruct the NGFS scenarios. Some parts of this article have been influenced by these conversations.