Climate Scenario Analysis 101

Climate change can create significant risk for banks, investors, and the economy. But how can regulators and firms measure and prepare for the possible financial ramifications of physical risks like hurricanes, heat waves, and wildfires, or transition risks created by the uncertain path towards net-zero? Increasingly, institutions are turning to climate scenario analysis (or, as it is referred to in some jurisdictions like the European Union, climate stress testing,) which is a process that institutions can undertake to assess their vulnerability to these climate risks. You can learn more about these climate risks in Climate Financial Risks 101.

The practice of climate scenario analysis is still in the early developmental stages: financial regulators in Europe, Australia, Canada, the United States, and elsewhere have started to conduct pilot exercises. This explainer provides an overview of how climate scenario analysis can be used to help financial institutions understand and prepare for the possible risks associated with climate change.

Stress testing as a concept is not new: introduced globally following the 2008 financial crisis, it is commonly used to analyze the ramifications of non-climate-related risks such as recessions. Many central banks, including the Federal Reserve and the European Central Bank, regularly conduct stress tests to assess banks’ ability to weather various market conditions.

Stress testing relies on simulation models. Modelers analyze how investments and their components—such as cash flow, debts, and returns—react to different macro-finance scenarios characterized by variables like inflation and fiscal policy. Hypothetical stressors can be generated in two ways: a re-creation of a historical stressor (such as the 2008 financial crisis) or a hypothetical event (such as a major war). The outcomes of these exercises can help determine whether institutions should adjust their operations, as well as how much capital a bank is required to keep on hand.

While many of these same factors play into climate scenario analysis, there are some notable differences. Existing analytical infrastructure can be used for climate scenario analysis, but some features need to be implemented or removed. Understanding these core differences and similarities will be a crucial part of ensuring that climate scenario analysis is accurate and useful while keeping the implementation burden manageable.

Below is a chart that represents the key differences between a conventional stress test and a climate scenario analysis.

Notably, climate scenario analysis features scenarios that have more dimensions than traditional stress tests: climate outcomes are determined by a mix of policy, geophysical events, and economics. Financial assets can also be exposed to the impacts of climate in nuanced ways: assets can encounter risks from natural disasters and their macro-finance feedback effects, as well as from the transition to new processes and energy sources.

The multifaceted nature of risks and impacts of climate change can make it difficult to accurately predict outcomes. Incorporating risks and impacts—especially when they can vary widely across geographies and sectors—requires granular data and models. The considerations, challenges, and outcomes of these more complicated models are outlined in the sections below. 

Climate scenario analysis differs significantly from conventional stress testing in multiple dimensions. Consequently, designing and implementing one requires additional considerations.

An important consideration concerns the design of scenarios that describe the economic and environmental contexts that potentially pose climate-related financial risks to business entities. These scenarios are not intended to be forecasts or probabilistic projections of likely climate and socio-economic trajectories (e.g., the RFF Socioeconomic Projections) and might feature an extreme case on purpose.

A plausible climate scenario requires consistency between its various components, including macroeconomic, energy, and climate variables. To capture the complex interplay between these components, one common approach is to use Integrated Assessment Models (IAMs) to model how these variables play out in different transition pathways. Major existing scenarios from the Network of Central Banks and Supervisors for Greening the Financial System (NGFS), the International Energy Agency (IEA), and the Intergovernmental Panel on Climate Change (IPCC) have all taken this approach.

As climate physical and transition risks are distinct in nature, geography, and sectoral concentration, modelers need separate assessments for each. Modelers must consider whether they will treat these risks as independent or dependent on each other; treating these risks as dependent makes the model more complicated but is likely more realistic. For example, when compared to a “business-as-usual” scenario, a scenario featuring strong climate policy will likely penalize high emitters, which would then change the rate of transition and physical risk based on myriad factors.

Existing exercises have taken different approaches on this matter. In the NGFS scenarios, the temperature variable is taken from its related transition pathways and then used to project damage to assets and productivity. As a result, the NGFS scenarios together depict a tradeoff between physical risks and transition risks. In contrast, the Federal Reserve’s pilot climate scenario analysis features independent scenarios for physical and transition risks, from the IPCC and NGFS, respectively.

A climate scenario analysis can be top-down or bottom-up A climate scenario analysis uses the terms “top-down” and “bottom-up” to describe who performs climate scenario analyses. These descriptions are not related to “top-down” and “bottom-up” modeling commonly used in economics research. depending on who performs the analysis. A top-down exercise is conducted by a regulator, whereas a bottom-up exercise is performed by individual banks and submitted to the regulator. A top-down approach ensures a consistent methodology and more comparable results across banks. A bottom-up approach allows for banks to individualize their climate risk assessment based on more detailed information, such as a more realistic representation of their risk management strategy and more granular data on those who takes out loans from them. However, this approach presents greater challenges to compare results across banks. Each bank might employ a different set of in-house models, assumptions, and data sources. Existing exercises show that banks often use estimated data from different sources, which creates large discrepancies in their input data even for the same counterparty. There is also a hybrid option, which combines top-down analysis with reports from individual banks.

A recent survey by the Financial Stability Board and NGFS finds that over half of the existing exercises are top-down, about one fifth are bottom-up, and the rest are hybrid. The prevalence of the top-down approach might be due to its relative simplicity and lower compliance burden on banks, which makes it more suitable as climate scenario analysis is in a nascent stage.

As regulators and banks build up their knowledge and capacity through the early exercises, they might move toward a hybrid or bottom-up approach to incorporate more granular data and realistic assumptions. Due to the challenges of the bottom-up approach described above, it will be crucial to establish best practices for documenting the key elements in the individual tests, which can inform regulators’ decision on the extent to unify the setup for the individual tests.

There are several areas of uncertainty to consider when modeling climate scenario analyses. It is unclear how adaptation and risk mitigation strategies such as new insurance policies, new technologies, and climate-smart management practices will be factored into climate scenario analyses. Past research suggests that some of these activities are highly effective: for example, having flood insurance largely eliminates debt defaults associated with flooding. However, it is difficult to systematically collect information about the type and extent of risk mitigation and risk transfer activities that will take place in different scenarios. Some activities depend on future technologies that might not be available yet, and their costs and risk-mitigating effects remain highly uncertain.

As different banks might make different assumptions, modeling these activities would introduce additional complexities to the findings. Given these challenges, some existing exercises have assumed a static balance sheet for banks, but that is also unrealistic. One possibility is to produce two sets of results, one with and one without assuming changes in banking behaviors in response to climate risks, which allows for scrutiny on the effect of these assumed changes. This is how the Federal Reserve treats insurance availability in two of the physical risk scenarios.

The choice of the time horizon also involves a key tradeoff. The full impacts of climate change, especially chronic physical risks, will manifest over a long time horizon (e.g., 30 years), beyond the typical time frame of a conventional stress test (1–10 years). However, a long time frame also compounds the deep uncertainties in the climate trajectory, as well as uncertainties underlying any assumptions made about model elements, particularly the potential response of the bank itself, making the output less informative.

Lastly, it is important to include secondary effects when evaluating the risk climate change poses to the financial system. Secondary effects might arise as climate risks are amplified and transmitted through the financial network, or when correlated risks become severe enough to trigger large-scale withdrawal of insurers (for example, if insurers stop offering fire coverage due to the prevalence of wildfires in a certain region). Existing exercises are highly limited in their treatment of secondary effects given a lack of understanding of these mechanisms or detailed data, which likely results in an underestimation of overall risk.

Climate scenario analyses are designed for regulators and banks to gauge the significance of climate risks on the financial health of both individual institutions and the overall financial system. They often use the same metrics of financial health as used in conventional stress testing, where standards of “passing” or “failing” the test are well established.  However, given the important limitations in the early climate scenario analyses described above, it is premature to draw conclusions about banks’ ability to withstand climate risks from existing findings. Thus far, regulators have treated these analyses mainly as informative exercises but have not imposed any real consequences on banks based on the findings, such as operational changes or increases in capital reserve requirements.

Beyond traditional financial metrics, climate scenario analyses can also generate additional climate exposure metrics that are useful for banks to measure and manage climate risks, such as the average flood risk of a mortgage portfolio or the average emissions intensity of a business loan portfolio. They can also be customized based on the specific climate policy context. For instance, the European Central Bank examines the emission-to-allowance gap, which measures firms’ exposure to carbon prices after accounting for free allowance under the European Union Emissions Trading System.

Special thanks to RFF scholars Billy Pizer and Marc Hafstead, as well as Adele Morris from the US Federal Reserve, for their invaluable input on this explainer.