Over the previous year and a half, regulators in the UK and EU have embarked on climate stress tests for financial institutions, following similar exercises lead by central banks in the Netherlands (DeNederlandscheBank) and France (Bank of France/ACPR). Most recently, the United States Federal Reserve Board announced a climate stress testing exercise in coordination with the nation’s 6 largest banks 4, and it seems plausible that a similar exercise could be expanded by other agencies that are part of the Financial Stability Oversight Council (FSOC). This blog aims to explore the different types of climate stress and scenario related exercises. As many are aware, climate stress and scenario modelling are imperfect exercises, replete with significant data challenges and requiring significant judgment for the assumptions used. However, they serve an important role as financial institutions start to grasp the extent of exposure to climate related risks in their portfolios and how best to manage them.
Shared Socioeconomic Pathways (SSP). SSP’s aim to merge social and economic models into a shared pathway. For example, the trajectory of carbon emissions created by economic growth depends on how that economic growth is generated. SSP’s account for population and economic growth, as well as technological and geopolitical assumptions and provide for green growth (SSP1), fossil fueled development (SSP5) and intermediate pathways. Each SSP corresponds to a global carbon emission pathway and concomitant radiative forcing that can be modelled to demonstrate potential physical climate impacts.
Representative Concentration Pathways (RCP). RCPs represent the evolution of carbon emissions in the atmosphere. The numeral next to each RCP corresponds to Global Warming Potential, a measure of radiative forcing caused by carbon and carbon equivalents in the atmosphere. RCPs are closely aligned with the SSPs, for example SSP5 (fossil fueled growth) and RCP 2.6 (orderly transition) are viewed as incompatible, absent significant assumptions on technological progress and commercial viability of carbon capture and storage (CCS), while SSP3 (Regional Rivalry) is incompatible with RCP 2.6 under all integrated assessment models.
Physical Risk. Fundamentally, physical risk is the exposure to the damaging effects of climate change on the planet – climate related events. Acute physical risks generally refer to tropical cyclones, wildfires, and other extreme events, while chronic physical risk refers to the gradual changes such as evolving precipitation patterns, sea level rise. Tipping points and feedback loops represent a significant gap in climate risk modelling as these are unpredictable, irreversible changes, such as the reversal of the gulf stream or the savannization of large areas of tropical rain forests.
Transition Risk. Transition risk is a firm’s exposure to the energy transition, with obvious impacts to firms in the fossil fuel and other carbon intense industries, but less obvious impacts to firms that are high dependent on carbon intensity through their value chains. Transition risk is most frequently measured through the price of carbon as a dependent variable that is proxied through assumptions of policy direction and Nationally Determined Contribution (NDC) declarations, adoption, and advances in renewable technology as well as changes in consumer preferences.
Regulatory Stress Tests
In 2021, following a COVID delay, the Bank of England launched the Climate Biennial Exploratory Scenario (CBES), with the stated objectives of protection of insurance holders, support of economic policy and the promotion of the safety and soundness of regulated entities and protection and enhancement of the stability of the financial system. The exercise was bifurcated into quantitative and qualitative components. The quantitative component considered 3 different pathways, consistent with the 3 major IPCC SSPs, Early Action (SSP 2.6), Delayed Action (SSP 4.5) and No Action (SSP 8.5). The qualitative component was a survey-based approach that gauged the risk governance and framework, consistent with PRA SS3/19. The scope of CBES was limited to the UK’s 19 largest insurers and banks. The key finding, with significant uncertainty, is that with no action on climate, climate risks could reduce profits by 10-15% on average, with significant idiosyncratic risks 1. Further, there could be a reduction of insurance coverage by 7% as properties become uninsurable.
In 2022, the ECB launched a similar exercise, with some material differences. While both the CBES and ECB modelled climate related risks consistent with the 3 shared socioeconomic pathways previously mentioned, the ECB relied on data submission of balance sheets while CBES looked to firms to perform the modelling. The ECB Stress test included both a qualitative survey of risk governance and framework of 104 participants and a quantitative modelling of 41 participants performed by the ECB. The results of climate modelling indicated that banks could have profit impacts in the order of 70 billion Euros under short term (3 year), disorderly scenarios 2.
The supervisory stress tests build upon the climate stress tests previous performed by the DeNederlandscheBank (2018) and the Bank of France (2019). While it is widely understood that insurers have exposures of lower duration with greater ability to exit markets and discontinue contracts with a risk profile inconsistent with the insurers risk appetite, insurers play an important role in risk mitigation for other financial institutions, as such, the evolution of insurance markets is an important consideration in the management of climate related risks for banks.
In the United States, the United States Federal Reserve has announced a climate stress test to be conducted with the nation’s 6 largest banks. The recent appointment of Nina Chen as the Chief Climate Risk Officer of the OCC3 as well as publication of draft principles on climate related financial risk management 5 lends more support to the view that climate risk management will be a high priority for US Regulators.
Approaches to Stress Testing
Scenario and Variable Selection
For climate scenario modelling, there are several decisions required at the initiation of the approach, as well as a need to establish governance and oversight for the process.
- Scenario Selection. Scenario selection involves selecting an expected future state of the world and modelling how the portfolio would perform. Climate risk modelling is unique in that it is forward looking and less reliant on historical data. While there are many published scenarios, there has been convergence on the importance of 4 key scenarios; RCP 2.6 (SSP1-2.6), RCP 4.5 (SSP2-4.5), RCP 6.0 (SSP4-6.0), and RCP 8.5 (SSP5-8.5), respectively, “orderly transition”, “disorderly transition”, “delayed transition” and “hot house world” (Note: notation used in IPCC AR6 is in parenthesis). Each of the SSP/RCPs noted previously represent a tradeoff between transition and physical risks, for simplicity, lower physical risks correspond to higher transition risks. Bespoke scenarios can be constructed that align with the firms view of the energy transition and evolution of physical risks, for example, a firm wanting to take a more conservative approach may wish to combine the higher transition risks pathway associated with SSP 1 or 2 with the higher physical risks associated with RCPs 6.0 or 8.5 pathway. Given the relative variability of forecasts, it is imperative to consider multiple scenarios and combinations of SSPs and RCPs.
- Time horizon. Climate scenario models are available through 2100, well beyond the horizon of even the highest duration portfolios, while traditional stress testing for risk management purposes largely relies on a 3–5-year horizon. Firms should consider the duration and liquidity of their portfolio when constructing climate stress tests to ensure relevance.
- Balance Sheet Dynamics. The balance sheet or portfolio of any financial institution is likely to evolve over time as firm strategy changes. The use of a static balance sheet provides for computational ease, but a more realistic balance sheet should reflect potential strategic pathways that account for firm, disinvestment, product, and net zero alignment strategy – as well as publicly announced commitments.
- Top Down, Bottom Up or Hybrid. Top-down approaches are the least granular stress tests, but also require the least data, effort and computationally intensity, while Bottom Up requires the most granular data, significant effort and computationally intensive. Hybrid Approaches bifurcates the portfolio into top down and bottom-up approaches.
Top-down stress testing model climate related impacts to key macroeconomic variables such as GDP, inflation, employment, sector, sub-sector, and market indices for each scenario chosen. The top-down approach is consistent with most active macroprudential stress tests such as CCAR, used by regulators to assess capital adequacy and financial stability. In 2018, the DNB of the Netherlands engaged in a pure top-down climate stress test that was exclusive to transition risk, across 3 banking groups and 29 insurers with a time horizon of 5 years and 4 scenarios. Climate modelled macroeconomic variables modelled for each scenario are decomposed into sector specific impacts, as different sectors would have different exposures and correlations to each climate scenario, for example, the insurance sector may have a stronger exposure to a physical risk heavy pathway, while carbon intense industries would have higher exposure to a transition risk heavy pathway. The final step is to quantify a financial impact by stressing the banks holdings against climate modelled macro-economic risk factors and variables. Macroeconomic, top-down, approaches may be sufficient for buy side firms predominantly concerned with market risk associated with liquid assets.
Bottom-up stress testing requires granular data about issuers, borrowers, lessees, and counterparties. To perform a bottom-up stress test, there is a need to disaggregate firms into sectors and subsectors as well as into specific revenue generating and go to market activities, supply chain vulnerabilities and the geospatial location of assets. Bottom-up stress testing aims to capture idiosyncratic or firm specific risks. For example, within the agricultural sector, a firm that has embraced drought resistant seeds, implemented approaches to more efficient use of water resources or has adopted lower emission technology, both machinery and fertilization, may have lower exposure to climate related risks from the energy transition as well as physical risks. Within the mining sector, firms may need to be disaggregated into revenues by specific minerals, as demand for rare earths are likely to increase, as are the costs and resources required for extraction of certain minerals. In addition to financial and geospatial considerations, climate mitigation and adaptation, as well as the credibility of firm’s management are considered. Bottom-up approaches are more appropriate for less liquid, long duration or highly concentrated issuer or sector exposures that cannot be easily removed from the balance sheet or portfolio. Bottom-up approaches may also be used by non-financial institutions to better manage operational risk, capital expenditure prioritization and net zero capital planning.
The hybrid approach combines aspects of both the bottom-up and top-down stress tests, largely a trade off between precision and resource consumption. The objective of the hybrid approach is to focus effort on sectors, borrowers or counterparties that have significant exposure to physical and transition risks, while moderating the data collection and computational effort required to model the full portfolio at that level of granularity. For example, a traditional commercial bank will likely have exposure to the professional services, financial institutions, technology, pharmaceutical, retail, mining, oil and gas, utility, agriculture, and real estate sectors. It seems logical to assume that technology, professional services, and financial institutions are likely to have lower exposure to physical and transition risks, while firms in the latter sectors may have higher climate risk exposures which require more granular risk management and proactive engagement from the financial institution.
Climate Scenario modelling is an essential, yet challenging exercise for financial institutions. While there is an impetus to wait on further regulatory guidance for climate scenario modelling, early adoption of climate scenario modelling will help to identify data and technology improvements, to upskill and identify talent, to implement operating model changes and to help management become better acquainted with the novel risks that are expected to proliferate and magnify in the coming months and years. The early adoption of climate scenario modelling may assist in gaining competitive advantage through the identification of new products, clients and markets and the avoidance of climate-related risks.
- Bank of England. Results of the 2021 Climate Biennial Exploratory Scenario, May 24, 2022 https://www.bankofengland.co.uk/stress-testing/2022/results-of-the-2021-climate-biennial-exploratory-scenario
- European Central Bank Banking Supervision. 2022 Climate Risk Stress Test, July 2022 2022 climate risk stress test (europa.eu)
- Office of the Comptroller of the Currency. OCC Announces Chief Climate Risk Officer, September 12th, 2022 OCC Announces Chief Climate Risk Officer | OCC
- Board of Governors of the Federal Reserve System. Federal Reserve Board announces that six of the nation’s largest banks will participate in a pilot climate scenario analysis exercise designed to enhance the ability of supervisors and firms to measure and manage climate-related financial risks September 29th 2022 https://www.federalreserve.gov/newsevents/pressreleases/other20220929a.htm
- Office of the Comptroller of the Currency. Risk Management: Principles for Climate-Related Financial Risk Management for Large Banks; Request for Feedback OCC Bulletin 2021-62| December 16, 2021 https://www.occ.gov/news-issuances/bulletins/2021/bulletin-2021-62.html