Climate Risk Stress Testing


Climate change is one of the most complex challenges that society faces today. The continued and growing emissions of greenhouse gases since the industrial revolution have led to increased GDP, but at the price of increasing temperatures creating risks to life, ecosystems, and economies. There is increasing consensus that risks posed by climate change may have a significant impact on the financial system. Because of this, growing efforts have been made to assess financial institutions’ exposures to climate related risks through climate risk stress tests.

Climate related risks fall into three broad categories: physical risks, transition risks and liability risks. Physical risks are those immediate threats that come from the physical environment. They arise from extreme weather events such as flooding, hurricanes, droughts, or wildfires. Transition risks are generated from the potential costs due to the introduction of policy, laws, and other regulations that are designed to adjust towards a low carbon economy. Transition risks also arise due to changes in technology and customer sentiment. The third risk category, liability risks, also come from climate change and transitioning. Liability risks are generated by a failure to mitigate, disclose, or comply with changing legal and regulatory expectations[1]. Currently, liability risks are not usually covered in climate risk stress testing exercises for financial institutions (Baudino and Svoronos, 2021).

Financial institutions are exposed to climate related risks directly through their operations. For instance, Banks may be affected by increasing costs of energy usage or their physical infrastructure, staff, customers, and vendors may be subject to extreme climate events. Indirectly, banks are exposed to climate related risks through the macroeconomic effects of disruptive transition pathways and extreme weather events. These economic effects can have an adverse impact on the rate of default on the loans and market prices of securities owned by financial institutions (Baudino and Svoronos, 2021). Currently, limited protection against these risks is provided by insurance contracts and such cover is unavailable in countries with less developed financial sectors (BIS, 2020.)

Since the 2015 Paris Agreement, initiatives have been undertaken to address climate risk within the financial sector, beginning with enhanced disclosure. In 2017, the G20 Financial Stability Board (FSB) launched the Task Force for Climate-Related Financial Disclosure (TCFD) which provides recommendations on disclosing climate change risks. In the same year, a group of central banks and financial supervisors established the Network for Greening the Financial System (NGFS). The NGFS has led growing efforts to incorporate climate-related risks within financial stability frameworks (NGFS, 2020). Additionally, the European Commission (EC) created the High-Level Expert Group on Sustainable Finance (HLEG) that recommended the introduction of standards for the identification of sustainable investments. These recommendations were included in the 2018 EC Action Plan for Sustainable Finance and guided the work of the EC Technical Expert Group on Sustainable Finance (TEG) (see European Commission, 2020) which culminated in the publication of the EU Taxonomy regulation in the Official Journal of the European Union in June 2020.

Zurich. “Here’s how climate change will impact businesses everywhere – and what can be done” https://www.zurich.com/en/knowledge/topics/climate-change/how-climate-change-will-impact-business-everywhere


These unprecedented initiatives are aimed at creating quantitative frameworks that assess risk exposure and potential losses due to climate change. Quantification of climate risk factors will provide key information for banks decision making for both financial and non-financial risks. Measurement is important as items that can’t be measured typically can’t be managed, planned, reported, and disclosed neither.

A key challenge that financial institutions face in quantifying climate related risks is a lack of relevant data. Traditional risk quantification techniques rely on past data using statistical risk modelling. However, unlike traditional risk analysis, there may be no data available for climate risks because transitional risks are unprecedented. Additionally, physical risk data may be irrelevant or unreliable as weather conditions are not consistent and instead are becoming more extreme and unpredictable. Data has also not been collected by financial institutions because climate change risks were not deemed to be material and so did not justify such efforts (Baudino and Svoronos, 2021).

To combat this challenge on the quantification of climate related risks, institutions are relying on stress testing. Stress tests are simulation exercises conducted to assess the resilience to a hypothetical scenario of either a single bank or the system (Baudino et al (2018)). Stress tests address the challenge of climate risks because they are forward looking. As such, they explore future exposures and potential losses that cannot be extrapolated from past data. They do not pretend to forecast which future scenario will actually happen, but rather describe the possible spectrum of future outcomes. Stress testing methodologies are familiar to banks because banks perform stress tests already today for risk management, planning and for regulatory compliance purposes.

In 2019 the NGFS recommended the use of climate stress tests for the assessment of the financial stability implications of climate risks (NGFS, 2019), and in 2020 they provided a set of standard climate scenarios that investors should consider in their climate financial risk assessments (NGFS, 2020). Stress testing scenarios and modelling techniques have been created by authorities and have been made publicly available to be employed across various jurisdictions.

The objective of this paper is to assess the design features of a climate risk stress test for banks using the NGFS framework. We shed light on the methodologies used by the NGFS and the main challenges associated with incorporating this framework. 


NGFS Scenario Design framework

The NGFS coordinated with climate scientists and economists to design a set of hypothetical scenarios that provide a reference point for understanding how climate change (physical risk) and climate policy and technology trends (transition risk) could evolve in different futures.


Within the NGFS framework, scenarios fall into four categories: (i) the “hot house world” category; (ii) the “too little, too late” category; (iii) the “orderly” transition category; and (iv) the “disorderly” transition category.

The NGFS has not designed specific scenarios for the “too little, too late” pathway. For each of the other three categories two granular scenario options are provided. These options reflect more the policy choices by the relevant authorities, technological change and the high-level of uncertainty in terms of outcomes, especially for physical risk.

 Orderly scenarios

Net zero 2050. Global warming is capped at 1.5°C through stringent climate policies and innovation. Global net zero CO2 emissions is reached around 2050. Some jurisdictions (e.g the United States, European Union and Japan) reach net zero for all GHGs.

Below 2°C. The stringency of climate policies increases gradually, so that there is a two thirds chance that global warming remains below 2°C.

Disorderly scenarios

Divergent net zero. Emissions reach net zero around 2050 but with higher costs due to divergent policies introduced across sectors leading to a quicker phase-out of oil use.

Delayed transition. This assumes annual emissions do not decrease until 2030. Strong and harsh policies are needed to limit warming to below 2°C. CO2 removal is limited. Physical risks and transition risks are higher because of the delay.

Hot house world scenarios (i.e current policies are implemented, but no additional measures are taken)

Nationally Determined Contributions (NDCs). This includes all pledged policies even if not yet implemented. Since these are insufficient to reach net zero by 2050, the increase in the mean temperature reaches 2.5°C by this date.

Current policies. This assumes that only currently implemented policies are implemented, and this is the “worst case” scenario. As a result, the increase in the mean temperature exceeds 3°C, with changes in the frequency and severity of severe weather events such as heatwaves, droughts, wildfires, tropical cyclones, and flooding. There are impacts on health, labour productivity, agriculture, ecosystems and sea levels.

Modelling Approaches

The transition pathways for the NGFS scenarios have been generated by well-established Integrated Assessment Models (IAMs), namely GCAM, MESSAGEix-GLOBIOM and REMIND-MAgPIE. IAM’s quantify the impact of climate change on macroeconomic variables. These models have been used in hundreds of peer-reviewed scientific studies on climate change mitigation. IAMs link the projection of GHG emissions to a corresponding set of carbon prices, energy demand and -prices as well as economic growth impacts. The IAM models provide GDP pathways, carbon prices, energy prices and -demand as well as GHG emissions for the scenarios. These variables are then used as inputs in macroeconomic models and applied to country blocks for regional results. The model frameworks are complemented by a sectoral model to assess economic impacts on specific sectors and by financial models to assess firm-specific and exposure-specific impacts (Baudino and Svoronos, 2021).

GCAM, MESSAGEix-GLOBIOM and REMIND-MAgPIE, share a similar structure. They combine macroeconomic, agriculture and land-use, energy, water and climate systems into a common numerical framework that enables the analysis of the non-linear dynamics in and between these components. Despite their similarities, each of the models has its own characteristics which can influence results. For instance, both MESSAGEix- GLOBIOM and REMIND-MAgPIE assume perfect foresight. This means participants in the model can fully anticipate changes occurring over the 21st century. In contrast, GCAM model has a “myopic” view of the future meaning participants consider only past and present circumstances in formulating their behaviour including expectations for the future. These differences affect the end results of the models (NGFS, 2021).


Climate stress testing by Central Banks provides them with important information on the overall financial system’s resilience to climate risks. For individual banks, stress testing offers a better understanding of sector exposures as well as their clients’ resilience and vulnerabilities. This measure is useful for guiding financial institutions in the adjustment of their risk management or rebalancing of their exposures. It provides management with information that could be useful for determining mitigating actions and adopting of transition plans. Banks are able to “green” their loan books by agreeing with their clients on binding transition plans and transition investments. Thus, climate change is an imminent risk but also a business opportunity for banks: climate change needs to be financed (by a.o. banks).

However, results from stress tests are often cautionary. According to the BIS (2021) disclosures about potential long-term losses could be questionable and possibly counterproductive, given the very high level of uncertainty surrounding the estimates and the risk that these could cause adverse reactions from market participants or even trigger market turmoil.


One of the main challenges faced in the conducting of climate risk stress tests is the lack of data. Firstly, data covering future climate patterns may be unavailable or unreliable, given the changes in climate patterns and weather conditions. Additionally, measuring the impact of climate risk requires granular data to which extent sectors and clients are exposed to climate risk. This information should ideally be categorised by sector and region to differentiate and assess risks along these dimensions. However, this data is often not available (Baudino and Svoronos, 2021).

Additionally, because climate risk stress testing is in its early stages there are not yet well established, common practices for financial institutions. While there is a consensus on the most carbon intensive emitting sectors, there are differences in methodologies in how carbon tax propagates through the economy. The use of different methodologies (like the three different simulation frameworks discussed earlier) limits comparability between stress testing exercise results. Thus, there may be benefits in developing commonalities.

Due to lacking data, methods, and standards, the implications of climate-risk stress tests are still “softer” than stressing traditional risks.

While traditional solvency risk stress tests have regulatory implications, climate risk stress tests implications on policy are yet to be defined. It is yet to be determined if climate stress tests are a suitable basis for quantitative requirements or whether they are better suited as a trigger for targeted discussions between supervisors and financial firms. some argue that the risks posed by climate change are too insignificant too substantially impact the financial system and climate financial regulation is a bid to pass laws that otherwise would not stand independently (Cochrane, 2021).  Despite this, banks are under rising regulatory and commercial pressure to protect themselves from the impact of climate change and to align with the global sustainability agenda. Therefore, the imperative to improve climate risk assessment and management is warranted.


Stress testing climate risks is relatively new and complex. Globally, financial authorities and banks are gradually building up capabilities. Stress test exercises, and the ways in which financial institutions act on their outcomes, can inform discussions on business models, internal governance, transition advisory and its risk management. Overall, banks need to develop long-term strategies for their business model to measure the risks of and finance the decarbonisation of economies. This will become a key priority in the near future.



  • Baudino P and Svoronos JP. 2021. Stress-testing banks for climate change – a comparison of practices.
  • Cochrane J. 2021. Climate Financial Risk. https://johnhcochrane.blogspot.com/2021/07/climate-risk-to-financial-system.html
  • Financial Stability Board (2020). The implications of climate change for financial stability, November.
  • Network for Greening the Financial System (2021). NGFS climate scenarios for central banks and supervisors.
  • Network for Greening the Financial System (2021). “Case Studies in Environmental Risk Methodologies”, NGFS Occasional Paper.