Friday, May 26, 2023
Introduction for Supervisors to Scenarios and Stress Tests of Climate Change Risks
This TC Note and accompanying podcast provide a detailed background on climate scenarios and practical guidance on the design of climate stress tests. Supervisors will gain an understanding of the uncertain nature of climate risks and the tools available to assess them.
Speaker:
R. Barry Johnston, Program Leader, Toronto Centre
Host:
Demet Çanakçı, Program Director, Toronto Centre
Read their biographies. Read the podcast transcript.
Listen to the Podcast:
Read the TCN:
INTRODUCTION FOR SUPERVISORS TO SCENARIOS AND STRESS TESTS OF CLIMATE CHANGE RISKS
TABLE OF CONTENTS
Climate change risks and the role of scenarios and stress tests
Financial risks posed by climate change
The future will be different from the past: limitations of statistical modelling
Use of scenarios in assessing climate risks
Technical considerations in designing climate scenarios
Use of stress testing in assessing climate risk exposures
Who should conduct the stress test?
Objective and scope of stress testing exercises
What climate scenarios should be used as the basis for designing stress tests?
What information and methodologies should be used?
What use should be made of the stress tests?
Example of a pilot stress test of financial risks from climate change
Copyright © Toronto Centre. All rights reserved.
Toronto Centre permits you to download, print, and use the content of this TC Note provided that: (i) such usage is not for any commercial purpose; (ii) you do not modify the content of this material; and (iii) you clearly and directly cite the content as belonging to the Toronto Centre.
Except as provided above, the contents of this TC Note may not be transmitted, transcribed, reproduced, stored or translated into any other form without the prior written permission of the Toronto Centre.
The information in this TC Note has been summarized and should not be regarded as complete or accurate in every detail.
INTRODUCTION FOR SUPERVISORS TO SCENARIOS AND STRESS TESTS OF CLIMATE CHANGE RISKS
Introduction1
Climate-related events and their associated risks are subject to significant uncertainty in their timing, frequency, and severity. Forward-looking assessment approaches are crucial to adequately account for the unprecedented nature of climate change. Against this backdrop, scenario analysis and stress testing are critical tools for assessing the potential implications of climate change on economies, financial institutions, and financial systems.
As part of their oversight of financial institutions, markets or instruments, and financial stability, supervisors need to understand the uncertain nature of climate risks, and the tools available to assess these risks.2 This Toronto Centre Note introduces financial sector supervisors to scenarios and stress testing of climate risks. The Note complements Toronto Centre (2020)3 on climate stress testing by providing detailed background on climate scenarios, and practical guidance on the design of climate stress tests.
The next section introduces the financial risks posed by climate change and explains the role of scenarios and stress testing in assessing those risks. The third section discusses the design of climate scenarios, including those developed by the Network for Greening the Financial System (NGFS). The fourth section examines the issues in the design of climate stress tests.
Climate change risks and the role of scenarios and stress tests
This section briefly reviews the types of financial risks posed by climate change and explains the role of scenario analysis and stress testing in identifying those risks.
Financial risks posed by climate change
As has been elaborated in other Toronto Centre Notes,4 climate change poses financial risks. In particular:
- Transition risks. These risks are associated with the transition to a lower carbon emissions economy, which would affect the value of assets, the viability of industries and sectors, and the prosperity of regions and national economies. Transition risks lead to concerns about so-called “stranded assets” – assets that would lose their value in the transition to a low-carbon economy. For example, a decision to phase out coal-fired energy generation because of its carbon footprint may result in some coal extraction becoming nonviable.
- Physical risks. These risks are associated with the damage caused by climate change to the physical environment and physical assets. An example is the impact of extreme weather events and sea level rise on higher insurance claims and lower collateral values in coastal communities.
- Systemic risks. These risks are associated with catastrophic tipping points due to potential non-linearities and feedback mechanisms. These risks have been characterized as “green swans”, with many of the characteristics of “black swans” that are rare and unpredictable events that can trigger systemic financial crises due to complex interactions.
The above risks create microfinancial and macrofinancial risks, as reflected in the following chart showing the transmission of climate risks to financial risks.
Source: NGFS (2021).
As they oversee financial institutions, markets or instruments, financial supervisors will normally be concerned with how well financial institutions identify and manage financial risks (illustrated in the box to the right in the above chart).5 Macroprudential authorities will be concerned with identifying and managing systemic risk associated with the feedbacks between the financial system, climate risks, and the macroeconomy.6
Traditionally, a range of techniques have been used to assess financial risks; many rely on historical data.7 However, climate risks are subject to:
- an unusual degree of uncertainty;
- future risks that are likely to be very different from the past; and
- information gaps and methodological challenges.
Two tools have become important to assess financial institutions’ exposures to climate change risks: scenario analysis and stress testing. The climate scenarios explore the implications of a range of possible climate paths for physical and transition risks and for economic variables. The scenarios help investigate both the uncertainty and the future paths for climate risks. The scenarios provide input on variables that are presented in the first two boxes in the above chart: the physical and transition risks and the impact on economic variables.
Financial risks (illustrated in the box to the right in the chart) will be specific to financial institutions and portfolios, determined by individual balance sheets and asset/liability exposures. To assess the financial risks, the scenarios need to be mapped to the individual balance sheets/asset/liability exposures. This is the role of stress tests. The preparation of stress tests also helps identify the information needs and gaps for assessing climate risks, and since the assessment of climate risks is relatively new, to develop methodological techniques to assess those risks.
The first major source of climate risk uncertainty is associated with the future path of greenhouse gas (GHG) emissions. The pace and severity of climate change is driven primarily by the world’s cumulative GHG emissions.8 While cumulative GHG emissions are already causing global warming, their future path, and hence the path for global warming and climate change, is highly uncertain. It will depend on such factors as:
- Public policy and the extent to which countries limit their GHG emissions to meet national and international climate objectives, including those set out in the 2015 Paris Climate Agreement (PCA) to keep global warming below two degrees centigrade. Countries make their climate commitments known in their Nationally Determined Commitments under the PCA. Public policy initiatives to limit GHG emissions include carbon taxes, subsidies for green (low-GHG-emitting) energy and technologies, including carbon capture, and laws banning or limiting brown (high-GHG-emitting) activities.
The global nature of emissions policy adds to the nature of climate uncertainty. Effectively mitigating emissions, and the physical risk of climate change, requires global action, which is the rationale behind the PCA and subsequent negotiations. Transition risks, however, generally reflect policies at the national level. Thus, a firm could confront extremely tough national emissions reductions, creating significant transition risk, while emissions overall continue a trajectory due to the inaction of other countries, leading to extreme physical risk.
- Technological advances in green energy production, including technologies for carbon capture. Such technological advances will change the economics of using brown compared with green production processes. Advances in technologies would mean that brown technologies would increasingly be replaced with green technologies, and that a given level of output could be achieved with lower (potentially much lower) GHG emissions. For example, solar and wind power are replacing coal-produced electricity as they have become cheaper.
- The level of economic activity. GHG emissions are closely associated with the overall level of economic activity. For example, GHG emissions fell in 2020 during the economic slowdown caused by the COVID-19 pandemic. The future path of GHG emissions will depend on the future level of economic growth, which itself will be impacted by climate change.
- Consumer and business preferences. Consumers and businesses may value green products over brown in their purchases and lifestyle/production and investment choices. Examples include consumer preferences to drive electric vehicles (EVs), and business decisions to source their energy needs from renewables rather than fossil fuels.
The above factors are not only important in driving the path for GHG emissions and global warming, but also for the speed of transition from brown to green assets. This also affects the timing and size of the transition risks, and over the longer term the magnitude of physical risks.
The second major source of climate risk uncertainty is associated with the impact of climate change on weather patterns and sea level rise that create physical risks. Extrapolating risks and losses associated with historical weather patterns may not accurately predict future weather-related risks and losses in the face of climate change.
The IPCC (2021) states that many changes in the climate system - including increases in the frequency and intensity of hot extremes, marine heatwaves, heavy precipitation, cyclones, and droughts - become larger in direct relation to global warming caused by concentrations of GHGs in the atmosphere. As global warming increases, chronic changes in climate may also amplify the impact of extreme events. For example, continued sea level rise may increase the typical levels of storm surge associated with a hurricane of a given intensity.
In addition, currently rare compound extreme events (separate extreme events affecting one location repeatedly or multiple locations simultaneously) may become more frequent, and there will be a higher likelihood of events with increased intensities, durations, frequencies, and/or spatial extents unprecedented in the observational record. Increasing physical risks are already becoming evident in many countries, and will intensify over time, but in ways that are hard to predict.
The third major source of uncertainty is associated with potential interdependencies and tipping points that amplify risk exposures because of the effects of climate change. Some of the interdependencies and tipping points result from interactions in the natural world, between higher global temperatures and factors impacting the major sources of physical risks, such as the speed of sea level rise. (These tipping points are discussed further below.)
Global warming is also a major cause of biodiversity loss, an additional source of physical and transition risk not explicitly accounted for in the current climate models. Other interdependencies are the result of policy actions, such as decisions to speed up the transition to a green economy because of intensifying physical risks. Still other interdependencies are the result of interconnections within the financial system. An example is the impact of climate change on coastal communities from the combination of sea level rise and exposure to more intense storms. If the exposure of these communities results in the withdrawal of property insurance coverage, this would transfer the risk of property losses from insurance companies to banks that have issued the household mortgages.
The future will be different from the past: limitations of statistical modelling
A critical tool in assessing financial risks is statistical modelling using historical data. Based on the assumption that the future will behave similarly to the past, historical data can be used to forecast future outcomes. Examples of statistical modelling include probabilities of default, expected losses given default, and value at risk. Statistical modelling underlies decisions on how much provisions to hold to cover expected losses on loans and capital to hold against unexpected losses.
An example of statistical modelling illustrating the considerations relevant to assessing the physical risks from climate change is the catastrophic loss (CAT) model used by (re)insurance companies. CAT models evolved in the 1980s because of:
- scientific progress understanding natural hazards and their meteorological, hydrological, climatological, and geological characteristics;
- engineering research and testing relating to the impact of hazards on the built environment; and
- progress with geographic information systems.
Traditionally, CAT models have relied on statistical techniques using empirical (observed) historical data of physical events.
CAT models generally involve four elements:
(1) a hazard module assessing the level of physical hazard across a region;
(2) an exposure module reflecting location within the region;
(3) a vulnerability module that estimates the percentage loss of the asset at risk; and
(4) a financial module that monetizes the losses from physical damage based on insurance policy terms and contract structures.
Source: The Geneva Association
Conceptually, the CAT framework for modelling losses is useful for evaluating the losses from the physical risks of climate change. However, its applicability is restricted by the model’s reliance on historical data to calibrate the losses, and by potential cascading effects and interdependencies of hazards and interaction between natural, technological, and critical infrastructure failures. Because the future risks of physical losses will be magnified by climate change, the use of historical data to calibrate the CAT models can significantly underestimate the financial losses.
Various proposals suggest ways to adapt the CAT modeling framework to the physical risks of climate change.9 Actuarial recommendations to assess the impact of climate risks emphasize the importance of using scenarios and stress testing.10
While statistical modelling techniques are the backbone of traditional risk analysis, their applicability to climate risk is limited. As the risks from climate change have barely started to materialize, standard approaches to modelling financial climate risk using statistical techniques will lead to the mispricing of risks.
Use of scenarios in assessing climate risks
Scenario analysis is designed to build understanding of future risks. Scenarios describe hypothetical future paths. They are not predictions or forecasts, such as might be generated by a statistical model based on historical data. Scenarios explore emerging risks in an uncertain future. Using a variety of scenarios can enhance critical thinking about the future. Scenarios support both qualitative and quantitative analyses of risks, including stress testing.
Before discussing the design of climate scenarios, it is useful to mention the technique of sensitivity analysis, which could be applied to climate risks. Sensitivity analysis examines the impact of single or multiple factors on the balance sheets of financial institutions.
- Sensitivity analysis: One or more moves in a particular risk factor, or a small number of risk factors that impact the balance sheet of the financial institution. For example, an assumed change in the exchange rate or interest rate.
- Scenario testing: Simultaneous moves in several risk factors impacting the balance sheets of financial institutions, linked to explicit changes in the view of the world.
An example of a sensitivity stress test would be to ask financial institutions to assess the impact on their portfolio of a one notch down grade in the credit ratings of their clients, or for banks to assess the impact of a one notch down grade in the quality of their loan portfolio.
Sensitivity analysis can be used to capture some features of climate-related risks. For example, the Bank of England (2019b) requested that insurance companies examine the impact of three broad categories of climate scenarios on their asset portfolios. The scenarios comprised:
A: sudden and disorderly transition;
B: progressive and orderly transition;
C: no transition.
For each of these scenarios, the Bank of England assigned a change in the equity value of investments in the fuel extraction and power generation sectors (see the following table) and asked the insurance companies to assess the impact on the asset side of their portfolios. The results would help to inform potential insurance company exposures to the transition risks from climate change.
Source: Bank of England (2019b)
Sensitivity analysis is useful for peer group analysis, as the same stress assumptions are applied across financial institutions. It is also useful to examine idiosyncratic shocks that fall outside statistical parameters. The shortcoming is that the design of the shock may be considered arbitrary, as it may not be based on a scientific model.
For jurisdictions with limited implementation capacities, sensitivity analysis may nevertheless provide a good starting point to examine climate risks. The size of the shocks for the sensitivity test could be derived from climate scenario databases, where information is available for the jurisdiction (for example, the NGFS scenarios discussed below); or lacking this, data from other jurisdictions facing similar climate risks.
The benefits of scenario analysis as applied to climate change are that scenarios can:
- Explore the most significant effects of climate change likely to emerge over medium- to longer-term time horizons;
- Explore the consequences of different potential paths and outcomes that reflect the uncertainty over the timing and magnitude of adjustments to climate change;
- Take into account potential complex interactions between changes in the climate and the economic environment in adapting to climate change; and
- Integrate climate science with macroeconomic and financial sector analysis. This allows for the generation of economic and financial sector variables (GDP, interest rates, exchange rates, sectoral activity, agricultural and labor productivity, etc) consistent with the climate projections.
The scenarios can be seen as the best scientific effort to understand and model the effects of climate change on national economies. They go beyond sensitivity analysis to provide the scientific basis to calibrate the shocks and generate a consistent set of shocks that can be used to stress-test financial institutions.
The Task Force on Climate-Related Financial Disclosures issued guidance on the use of scenario analysis in the disclosure of climate related financial risks.11 They outlined five desirable elements in designing climate scenarios, namely that the scenarios should be:
- Plausible – the events described in the scenarios are possible and the narratives are credible.
- Consistent – there is a strong internal logic in the scenarios.
- Relevant – the scenarios should contribute insights into the implications of climate change on the natural and economic environment.
- Challenging – the scenarios should challenge conventional wisdom and simplistic assumptions about the future (such as simple extrapolation of historical trends).
- Distinctive – the scenarios should allow for different combinations of key factors so as to explore the range of possible outcomes from climate change.
Plausibility and consistency in the scenarios can be established by building models to represent the economic, social, and natural environment. The model frameworks should reflect a consistent understanding of the functioning of the natural, social, and economic environments, with a strong internal logic.
Models are necessarily abstractions with many possible model designs. What is important is that the models are “fit for purpose” in both theory and practice, and have been thoroughly vetted in the scientific community, so the results generated (even if extreme) will be plausible.
The relevant and challenging characteristics of the scenarios can be introduced through the parameters and assumptions used in the simulation; for example, the path for GHG emissions. Distinctive features can be introduced as part of model design to explore different aspects of the effects of climate change and different interactions. The different models should include explanations of their significant limitations, since various trade-offs are inevitable in design.
Technical considerations in designing climate scenarios
Climate scenarios combine knowledge of the atmospheric and natural environment with that of the social and economic environment to project the impact of climate change on economic activity and exposures to transition and physical risks. The technical analysis provides a range of possible outcomes, which need to be prioritized for the purpose of implementing climate stress tests (see below).
Developing the scenarios involves various building blocks. The common starting point in analyzing physical and transition risks are climate projections. Projections of GHG emissions determine different [representative] climate pathways (RCPs, for example those derived from the work of the Intergovernmental Panel on Climate Change (IPCC 2021)). These GHG concentration trajectories represent a widely referenced set of projections about the range of possible governmental policies and socioeconomic trends developed with input from domestic and international climate experts. The IPCC climate projections are the starting point for most climate scenario analysis.
The IPCC identified four key temperature pathways for future GHG emissions. These represent a reasonable range of possible future states:
- RCP 2.6 is consistent with an ambitious reduction in emissions to limit global warming to less than two degrees Celsius (2°C) above pre-industrial levels, the goal of the Paris Agreement.
- RCP 4.5 is an intermediate emissions scenario. Emissions would increase modestly until 2040 before declining. It is likely to produce warming of about 2.4°C.
- RCP 6.0 is a high-intermediate scenario, where emissions peak around 2060 and decline thereafter. It is likely to produce warming of about 2.8°C.
- RCP 8.5 is a scenario assuming little action to reduce emissions. It is likely to produce warming of about 4.3°C. While extreme, it is not intended to represent a worst-case scenario.
Physical risks are often modelled using damage functions that relate the RCPs to losses from physical damage. The projections of physical damage usually distinguish between:
- Acute impacts from extreme weather events, which can lead to business disruption and damages to property and infrastructure, increase underwriting risks for insurers, and impaired asset values; and
- Chronic impacts, particularly from increased temperatures, sea level rise, and precipitation, affecting labor, capital, land, and natural capital These changes will require a significant level of investment and adaptation from companies, households, and governments.
The physical risks for specific jurisdictions are derived from projections of the weather patterns impacting those jurisdictions under the different RCPs.
The projections of transitions risks are typically modelled using Integrated Assessment Models (IAMs) that allow for the interactions between economic activity and emissions. IAMs combine macroeconomic, agriculture and land-use, energy, water, technological advances, and climate systems into a common numerical framework that enables the analysis of the complex and non-linear dynamics between these components. IAMs rely on calibration rather than econometric (historical data) estimation.12
A key variable linking the RCPs to economic variables in IAMs is the shadow price of carbon. More ambitious climate pathways imply lower GHG emissions and a higher shadow price of carbon. The advantage of IAMs is that they can calculate the shadow price of carbon along a reference path of output, emissions, and climate change. The shadow price of carbon is the key variable that drives the transition to a lower-carbon economy in the models and generates the transition risks in the scenarios.
Macroeconomic models use the output from the IAMs, including the shadow price of carbon, to generate projections for economic variables, GDP, interest rates, etc. The macroeconomic models may be designed at the national or regional level.
The scenario outputs generated through the above processes are model specific: the RCPs can lead to different projections of physical and transition risks and economic impacts depending on model design. For this reason, multiple scenario outputs may be generated for each RCP to account for different model designs and to reflect model uncertainty.
The NGFS developed its climate scenarios to provide central banks and supervisors with a common starting point for analyzing climate risks under different future climate pathways. The NGFS scenarios reflect different combinations of economic, technological, and policy assumptions that generate projections for economic and financial variables like GDP growth and carbon prices.13 Many financial authorities have used or adapted the NGFS scenarios for their climate scenario exercises.14
The NGFS scenarios represent different levels of physical and transition risks. These scenarios are neither forecasts nor policy prescriptions and do not necessarily represent the most likely future outcomes or a comprehensive set of possible outcomes. Rather, they represent a range of plausible future outcomes that can help build understanding of how certain climate-related financial risks could materialise, and how these risks may differ from the past.
NGFS distinguishes three main scenarios:
- Orderly: Orderly scenarios assume climate policies are introduced early and become gradually more stringent. Both physical and transition risks are relatively subdued.
- Disorderly: Disorderly scenarios explore higher transition risk due to policies being delayed or divergent across countries and sectors. For example, carbon prices are typically higher for a given temperature outcome.
- Hot House World: Hot House World scenarios assume that some climate policies are implemented in some jurisdictions, but efforts are insufficient globally to halt significant global warming. The scenarios result in severe physical risk, including irreversible impacts like sea-level rise.
The three main scenarios are further disaggregated (see graphic below). The implications of the different scenarios for the size of physical and transition risks are shown in the graphic.
Source: NGFS (2021).
Each NGFS scenario explores a different set of assumptions for how climate policy, emissions, and temperatures evolve:
- Net Zero 2050 limits global warming to 1.5°C through stringent climate policies and innovation, reaching global net zero CO2 emissions around 2050.
- Below 2°C gradually increases the stringency of climate policies, giving a 67% chance of limiting global warming to below 2°C.
- Divergent Net Zero reaches 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 assumes annual emissions do not decrease until 2030. Strong policies are needed to limit warming to below 2°C. Negative emissions are limited.
- Nationally Determined Commitments include all pledged targets even if not yet backed up by implemented effective policies.
- Current Policies assumes that only currently implemented policies are preserved, leading to high physical risks.
The Current Policies scenario is the most adverse in terms of physical risks, while the Net Zero 2050 scenario reflects a relatively smooth transition to net zero emissions by 2050. In the Delayed Transition scenario, emissions are only reduced after 2030, and hence require more rapid adjustments to limit the most severe physical impacts resulting in high transition risks.
The NGFS produces macroeconomic, financial, transition variables, and physical risk factors consistent with each scenario. These variables are available in the NGFS Scenarios Database hosted by the International Institute for Applied Systems Analysis and are available through the NGFS Scenarios Portal (2023).
The exposures to physical risks are found on the NGFS Climate Impact Explorer,15 which provides data on a range of physical risks, such as exposures to extreme weather and impacts on agricultural yields, over different time horizons. The information is available for individual jurisdictions and in larger countries for sub regions. Transition and economic variables consistent with each NGFS scenario are available in the NGFS Scenarios Database.16 Detailed information on a range of economic and transition variables by jurisdiction is available for download.
The earth’s climate is a complex, nonlinear system. Highly nonlinear systems can lead to chaotic dynamics, which are extremely difficult to model with any accuracy and confidence. As global warming continues, the world faces a situation of deep uncertainty related to the biogeochemical processes that can be triggered by climate change.
Various potential tipping points could dramatically worsen the effects of global warming (see graphic below). In the graphic, the individual tipping elements are colour-coded according to estimated thresholds in global average surface temperature. Arrows show the potential interactions that could generate cascades, based on expert assessment. Some potential tipping cascades are more likely to occur if there is global warming of between 1°C and 3°C, while others are more likely to occur if global warming exceeds 3°C or 5°C. Many tipping points may occur even if the world manages to keep global warming below 2°C.