Industry
Banking
Banking Services Transformation: Unlocking Exponential Value
Highlights
- Current climate stress tests rely on linear assumptions and legacy models, causing capital impacts to appear minimal and masking the true scale of systemic climate vulnerabilities.
- Modeling frameworks often omit complex feedback loops and cascading disruptions — such as geopolitical shifts, supply‑chain shocks, and resource scarcity — creating a misleading impression of low risk.
- Stress testing needs a structural overhaul, with scenarios that incorporate nonlinear impacts, second order effects, and high impact tail events to better inform capital planning and strategic decisions.
On this page
On this pageinpage
Rethink climate risk
Despite years of mounting scientific evidence and increasingly visible climate impacts, climate risk remains largely underrepresented where it matters most: in stress testing outputs and capital planning. Corporate and banking disclosures have grown increasingly sophisticated, with extensive climate-related metrics, targets, and transition plans. Yet, for all this visibility, the capital impacts derived from regulatory and internal stress tests often remain marginal or even negligible. This is likely to create the impression that climate risk is manageable or limited in scope.
This article suggests that current assumptions in place — particularly around time horizons, materiality, and systemic interactions — need revisiting to ensure that climate risk is better captured in stress testing. While disclosure frameworks have evolved, stress testing methods are yet to catch up, remaining constrained by legacy paradigms that often struggle to reflect the nonlinear and systemic nature of climate risk.
High-level outlook
Many current stress testing approaches integrate climate variables into legacy financial risk models. This often results to a less than complete picture of risk. Most scenarios assume gradual, linear transitions, orderly market responses, and limited second-order effects. Institutions may follow regulatory guidance precisely, but the resulting outputs typically show modest losses, low capital impact, and minimal implications for risk appetite or strategic planning.
This technical compliance can provide comfort — but it may not reflect the true nature of the risk. Tools designed for cyclical or idiosyncratic risks are being applied to a structural, systemic challenge. As long as models use assumptions that downplay tail risks, omit feedback loops, and truncate long-term consequences, stress test outputs will likely continue to show reassuringly low results — results that may not fully represent underlying vulnerabilities.
Integrated Assessment Models (IAMs), for example, typically connect climate pathways to macroeconomic outcomes through a narrow set of transmission channels, such as energy or carbon pricing effects. These models may overlook more complex feedback loops, social disruptions, or geopolitical shifts. When used as inputs into stress testing frameworks without adaptation for financial relevance, they can produce an overly simplified view of transition risk.
This reflects a broader modelling paradox: models often tend to focus on what can be quantified rather than what is most important. Elements that are complex, hard to quantify, or poorly understood tend to be excluded — not because they are unimportant, but because they do not fit neatly into existing frameworks. Their exclusion, however, can result in outputs that understate risk exposure.
The recent Network for Greening the Financial System (NGFS) release of short-term climate scenarios illustrates this challenge. While an important step forward, these scenarios still rely heavily on IAMs and link climate risk to macroeconomic outcomes primarily through energy price shocks. This limited transmission mechanism constrains the range of potential disruptions reflected in the results, reinforcing the perception of manageability.
Low risk illusion
Consider the situation where stress testing results indicate low risk: small capital impact, no trigger for strategy changes. One might be asked: “Can this be taken as confirmation that the risk is minor?” Based purely on the numbers, it can be tempting to answer “yes.” But such an answer may lead to misrepresenting the reality.
Low capital impact in a climate stress test does not prove low climate risk. It suggests the scenario may not have captured systemic pathways, abrupt shifts, or compound effects. Metrics only reflect what they are designed to measure — but in our view, those designs need to be broadened.
Comprehensive assessments may point to what is often excluded: cascading effects of resource shortages, widespread insurance market withdrawals, geopolitical instability, or the breakdown of systems that support economic stability. If these factors remain outside the scope of stress testing, the resulting outputs may convey a sense of security not because the risks are minimal, but because the models are constrained.
Nonlinear impacts
Figure 1 - Interconnected Drivers of Nonlinear Climate Risk
Stress testing methodologies have remained largely linear and siloed. They often rely on gradual scenario paths, static balance sheets, and sector-specific shocks. Yet climate impacts can cascade, amplify, and interact in complex ways (see Figure 1).
For example, a drought may reduce agricultural output but can also disrupt supply chains, increase commodity prices, destabilize governments, and affect regional creditworthiness. Such cross-sectoral and transboundary interconnections are typically underrepresented in existing frameworks.
That is one of the reasons capital impacts from climate stress tests may appear negligible – the models need to be better equipped to reflect the full scale and complexity of potential outcomes.
Stress tests rewired
Making climate risk more visible through stress testing requires rethinking how these tests are designed. This involves modelling interactions between physical and transition risks, across sectors, geographies, and time horizons. It calls for revisiting assumptions about mean reversion, market stability, and isolated risk factors.
Stress tests should ideally be capable of reflecting the potential for cascading impacts: infrastructure loss leading to credit deterioration, resource scarcity triggering social unrest, or market repricing of stranded assets cascading through portfolios.
The COVID-19 pandemic demonstrated how a single systemic shock can ripple through every layer of the economy. Climate change will not be a single event, but its impacts may be broader, more frequent, and more persistent. And that is precisely why there is a need for adaptation – without which stress testing frameworks may not be able to accurately predict and identify these vulnerabilities.
Redefining scenarios
Scenario analysis is a right tool – but it needs to evolve. To better illuminate climate risk through stress tests, scenario design should go beyond regulatory minimums and aim to stress underlying assumptions meaningfully. This includes embedding systemic logic, modelling second-order effects, and considering low-likelihood but high-impact events.
Importantly, such scenarios should inform capital planning, risk appetite, and business strategy. TCS can support clients in examining embedded assumptions and designing bespoke, high-impact narratives tailored to their portfolios and operations, working with scenario development, capital, and risk teams to translate these into quantifiable, actionable insights.
Closing the gap
There has been notable progress in climate risk reporting across the financial sector. However, a key challenge remains: these disclosures do not always translate into stress testing results that inform capital planning or strategic decision-making. This disconnect often reflects a misalignment between current stress testing tools and the characteristics of climate-related risks.
Relying on legacy approaches to model future shocks may limit the ability to capture emerging vulnerabilities. Reassessing stress testing frameworks — including their underlying assumptions, models, and scenario design — can help ensure they are better aligned with the evolving risk landscape.