NGFS phase V climate risk scenarios: good progress however unaddressed limitations remain

In November 2024, the NGFS[1] (Network for Greening the Financial System) published phase V of its widely used climate scenarios updated with the most recent economic and climate data, policy commitments, and model versions. This phase V also introduces a new damage function for physical risk assessment.

The new damage function incorporates various climate variables and persistence effects to capture diverse impacts on economic output. It uses high-resolution datasets for global applicability and includes lagged variables to account for delayed economic consequences. While it offers enhanced accuracy and realistic dynamics, it has limitations such as uncertainty in projections and potential overfitting. The function highlights the significant impact of physical risks on economic outcomes, including gross domestic product (GDP), inflation, unemployment, and interest rates, and influences the valuation of various asset classes.

The new NGFS scenarios align with advanced climate research, providing detailed projections of economic losses. They aid in analysing transition and physical risks for financial stability but need adaptation to specific contexts. Collaboration among financial institutions and regulators will be crucial, though reliance on shared models may create blind spots, requiring alternative scenarios and independent validations.

In this article we will explore the following in further detail:

1. Introduction to the NGFS climate scenarios
2. Phase V – details and analysis
3. Implications for financial institutions

1)  Introduction to the NGFS climate scenarios

In November 2024, the NGFS published phase V of its widely used climate scenarios.

The NGFS is a group of central banks and financial supervisors aiming to share best practices and mobilise finance for a sustainable economy. Every year they publish climate risk scenarios that regulators and financial institutions use for climate stress testing (CST) purposes among others. Details of the five phases are provided in the diagram below (Figure 2).

The NGFS long-term climate scenarios outline potential economic developments under various assumptions. Designed as a common framework, there are seven scenarios that help to analyse the impact of climate risks on the economy and financial system. They explore a range of plausible futures, rather than making forecasts, and produce results that integrate transition and physical risks with macro-financial developments. The scenarios are based around four dimensions: Orderly[2], Disorderly[3], Hot House World[4], and Too-Little-Too-Late[5] (details provided in Figure 1).

The NGFS regularly updates its scenarios to integrate the latest climate science, the progress of countries in adapting to and mitigating climate risks (including updated NDC[6]), and user feedback. These updates enhance the accuracy and usability of the scenarios for regulatory purposes.

Adoption of NGFS scenarios in regulatory CST: 

There has been widespread adoption[7] of NGFS scenarios for national and regional regulatory climate risk stress tests. This widespread adoption helps harmonise climate risk assessments across different jurisdictions and helps regulatory authorities better understand and manage climate-related financial risks. Regulatory authorities using NGFS scenarios in their stress testing frameworks include:

• NGFS phase II: Bank of England[8], European Central Bank[9], and the Bank of Canada[10].
• NGFS phase III: The Monetary Authority of Singapore for their 2023 industry-wide stress test exercise[11].
• NGFS phase IV: Hong Kong Monetary Authority[12] and Bank of Japan[13].

2)  Phase V – details and analysis

Key updates:

Introduction of a new damage function

The latest scenarios feature an update to the modelling framework to incorporate a new damage function developed by Kotz, Levermann, & Wenz, 2024[14]. This feature enhances the assessment of physical climate risks by integrating a broader range of climatic variables and capturing the long-term impacts of climate shocks on economic output.

New models and updated policy commitments

There are new updates and enhancements in line with evolving expectations, models, and policy commitments. Namely:

• An updated version of the Shared-Socioeconomic Pathways (SSPs) – The use of the SSP2 v3.0[15] as the baseline for all scenarios in phase V means that the scenarios are built on a consistent ‘middle-of-the-road’ assumption about global economic, social, and demographic trends. This ensures that all scenarios are comparable and makes it easier to isolate the effects on the economy because of climate policy decisions and physical risks.
• Incorporating the latest policy commitments under the United Nations Framework Convention on Climate Change (UNFCCC)[16].
• Refining of carbon dioxide removal (CDR) technologies including the deployment assumptions, cost and feasibility and the potential limitations. Previous iterations of the NGFS scenarios relied on optimistic assumptions about CDR technologies. The refinement ensures that the scenarios reflect real-world conditions and potential risks more accurately, reducing the risk of relying on overly optimistic technologies.
• Structural updates to two of the Integrated Assessment Models (IAM) models to enhance their functionality and alignment with updated climate and economic assumptions. The updates were made to the REMIND_MAgPIE and MESSAGEix-GLOBIOM models.
  • REMIND_MAgPIE: Enhanced representation of CDR technologies, representation of land-use transitions and energy system feedback[17].
  • MESSAGEix-GLOBIOM: To enable the analysis of economic impacts of delayed or partial climate policy implementations at a more detailed sectoral breakdown was introduced[18].

Higher peak temperatures across all scenarios

As a result of the slow implementation of climate policies by country, higher emission levels were observed in the near-term. The peak temperatures reached under the orderly scenario are now higher compared to phase IV and have become more disorderly leading to higher carbon prices in the short term compared to phase IV[19].

Technical details on the phase V updates:

Damage function

Key assumptions:

• Expanded climate variables: The new damage function incorporates variables beyond mean temperature, such as temperature variability, precipitation patterns, number of wet days and rainfall extremes, to capture diverse climate impacts.
• Persistence effects: It accounts for delayed economic consequences of climate shocks, using lagged climate variables
• Level effects framework: The model assumes that climate shocks affect the level of economic output rather than permanently altering growth rates. However, lagged variables ensure that medium-term impacts are considered.
• Robust calibration: The function leverages high-resolution datasets from the ISIMIP[20] and CMIP-6[21] models, ensuring global applicability and granularity in its projections.

Strengths:

• Enhanced accuracy: By integrating multiple weather variables and persistence effects, the function offers a more comprehensive view of climate impacts.
• Applicability: The use of detailed historical and projected economic data enables global and regional analyses.
• Realistic dynamics: Unlike static models, the inclusion of lags captures the gradual recovery of economies post-climate shocks.

Limitations:

The new damage function is a key improvement for integrating climate risks into economic models, although it comes with several limitations:

• Uncertainty in projections: The model’s reliance on median scenarios and confidence intervals highlights inherent uncertainties in long-term climate predictions.
• Exclusion of some factors: The damage function remains the same as its predecessor given the consistency in the incomplete modelling of acute climate risks. Acute risks like cyclones and long-term phenomena such as sea level rise are only partially captured, potentially underestimating total damages.
• Potential overfitting: The inclusion of numerous variables and lags increases the risk of overfitting, which could affect out-of-sample predictions.
• Simplified assumptions: Assumptions like unweighted regression could skew results, especially for regions with limited data coverage. The regression model only uses the first-order differences of GDP and climate variables. The choice of this specification is attributed to the fact that most of these variables are first-order stationary. But this is not a rule, as instances of non-stationarity could occur, thus trumping result interpretability.

Implications for interpretation:

While this new damage function marks a significant improvement in modelling physical risks, users must approach its results with caution. The projections should not be interpreted as standalone forecasts but as part of a broader analytical framework that includes socio-economic pathways, adaptation strategies, and acute risk assessments. Complementary tools and analyses are essential to fully capture the multifaceted nature of climate impacts on global economies.

Impact on physical and transition risks

The updates made to the scenarios under NGFS phase V have resulted in higher physical risk estimates and higher carbon prices required for an orderly transition. In all scenarios, it was observed that the impact of physical risks outweighs the impact of transition effects. The implementation of the new damage function has resulted in a fourfold increase in the impact of physical risks by 2025 in some scenarios.

Economic outcomes

• GDP: GDP losses are higher where mitigation actions are delayed. Scenarios with weak or no additional climate risk policy result in significant projected GDP losses – GDP can range from 5% to 15% by 2050. The GDP losses are generally lower by 2050 if transitioning to net zero by 2050 occurs – the estimated losses can range from 2% to 7% by 2050[22].
• Inflation: Transitioning to a low-carbon economy increases inflation in the short term after which it stabilises, however, if the action is delayed the inflationary pressure rises due to transition costs.
• Unemployment: Job creation in green sectors can offset some of the employment losses in fossil fuel-dependent industries, but if the transition is not well managed there may be a spike in the unemployment rate due to sectoral distribution.
• Interest rates: Interest rates tend to remain stable in the case of a more gradual transition, while higher risks and greater uncertainty in delayed transition scenarios could push up interest rates.

Impact on various asset classes

The updates in the NGFS phase V influence the valuation of different asset classes over time[23].

• Bonds: Sovereign bonds may experience higher yields and downgrades due to climate change and growing vulnerability to physical risks. Corporate bonds in high-carbon industries may face increased borrowing costs compared to green bonds (which fund renewable projects).
• Real estate: Investments in sustainable real estate are expected to increase. Properties in higher-risk zones may lose value or face higher insurance premiums.
• Private equity: Green investments that focus on clean technology and sustainable infrastructure will likely see increased investment, while high-carbon assets may experience declining valuations if they do not transition to low-carbon strategies.
• Commodities: an increase in the demand for renewable energy may lead to a decrease in the demand for fossil fuels leading to price volatility of fossil fuels like oil, gas and coal.
• Currencies: Countries that implement climate policies are more likely to see an appreciation in their currency compared to those lagging behind.
• Equities: Companies in high-carbon sectors face significant risk and may experience lower valuations compared to companies in green sectors because of regulatory pressure, carbon pricing, etc.

3)  Implications for financial institutions

Both significant and less-significant institutions are expected by central banks and supervisors to increasingly integrate NGFS scenarios into their risk management frameworks. This new phase aligns the NGFS scenarios with state-of-the-art climate-related research, providing more robust and granular projections of economic losses under various climate conditions. The damage function highlights the NGFS’s dedication to refining its climate scenarios and emphasises the importance of ongoing research to tackle the complexity of climate risk. The enhanced models deliver critical insights for policymakers and financial institutions to design effective risk management strategies in an era of accelerating climate change.

Financial stability, adaptation and limitations

NGFS scenarios provide a common framework for analysing both transition and physical risks, essential for regulatory and supervisory purposes. By using these scenarios to conduct risk analyses and guide financial stability efforts, regulators can help financial institutions navigate climate-related risks and ensure a resilient financial system against both transition and physical risks. In return, financial institutions must analyse the impacts of these scenarios on their businesses and report to regulators.

While NGFS scenarios are comprehensive, they need to be adapted by users to fit specific contexts and objectives as they do not capture every possible implication of climate change. Users are encouraged to supplement them with additional tools and data to address, for example, societal impacts, compound risks[24] and institutions’ asset exposure specificities. The Bank of England’s April article[25] on scenario analysis recommendations highlight this need.

The previous ECB CST 2022 was based on the NGFS scenarios and the recent European Supervisory Authorities’ and European Central Bank’s (ECB) joint Fit-for-55 climate risk scenario analysis[26] also used NGFS phase IV scenarios. As such, we can expect the next regulatory CST exercise to be based on phase V NGFS scenarios.

Integration and collaboration

The integration of NGFS scenarios into risk management frameworks involves conducting scenario analyses and disclosing climate-related risks in line with regulatory expectations. Increased collaboration among financial institutions, regulators, and stakeholders can foster a sustainable financial system but may lead to shared blind spots if they rely on the same scenarios and assumptions. To mitigate this, they should engage in developing alternative scenarios or complementary frameworks alongside the NGFS ones, independent validations, scenario expansion, and regulatory oversight to monitor systematic risks from herd behaviour or overreliance on shared models. This collective effort will also help the NGFS to adapt their scenarios over time with more detailed methodologies and tailored scenarios to support the transition to net-zero.

[1] NGFS Website: NGFS; [2] Orderly scenarios assume climate policies are introduced early and become gradually more stringent. Both physical and transition risks are relatively subdued.; [3] Disorderly scenarios explore higher transition risks due to policies being delayed or divergent across countries and sectors. For example, (shadow) carbon prices are typically higher for a given temperature outcome.; [4] Hot house world scenarios assume that some climate policies are implemented in some jurisdictions, but globally efforts are insufficient to halt significant global warming. The scenarios result in severe physical risk including irreversible impacts.; [5] Too-little-too-late scenarios assume that a late and uncoordinated transition fails to limit physical risks.; [6] Nationally Determined Contribution represent the climate actions outlined by countries to achieve the long-term goals of the Paris Agreement (more information here: Nationally Determined Contributions (NDCs) | UNFCCC); [7] NGFS publication: Scenarios in action – A progress report on global supervisory and central bank climate scenario exercises; [8] Bank of England – 2021 Climate Biennial Exploratory Scenario: Results of the 2021 Climate Biennial Exploratory Scenario (CBES) | Bank of England; [9] European Central Bank – 2022 Climate Stress Test: 2022 climate risk stress test; [10] Bank of Canada and the Office of the Superintendent of Financial Institutions – 2021 Pilot Climate scenarios Analysis: Using Scenario Analysis to Assess Climate Transition Risk; [11] MAS – Regulatory Updates and Expectations on Appointed & Certifying Actuaries; [12] Hong Kong Monetary Authority – 2023 Climate Risk Stress-Test: Guidelines for Banking Sector Climate Risk Stress Test; [13] Bank of Japan – 2022 Pilot Climate Scenario Analysis: BoJ – Pilot Scenario Analysis Exercise on Climate-Related (August 2022); [14] The economic commitment of climate change | Nature; [15] SSP Scenario Explorer (SSP 3.0, Release January 2024); [16] Nationally Determined Contributions Registry | UNFCCC and Policies | Climate Policy Database; [17] NGFS Scenarios – Technical Documentation – Phase V (pages 41-59); [18] NGFS Scenarios – Technical Documentation – Phase V (pages 60-82); [19] NGFS Scenarios – Main Presentation – Phase V; [20] Inter-Sectoral Impact Model Intercomparison Project studies the potential impacts of climate change under different climate-change scenarios.; [21] Coupled Model Intercomparison Project Phase 6 is a climate modelling project providing detailed projections and insights into how the Earth’s climate system responds to various forcing scenarios.; [22] NGFS Scenarios Portal; [23] NGFS Phase 5 Scenario Explorer; [24] The combination of various climate-related events (cf. NGFS-Compound risks implications for physical climate scenario analysis); [25] The Bank of England shares useful insights to measure climate-related financial risks using scenario analysis – Forvis Mazars – United Kingdom; [26] Transition risk losses alone unlikely to threaten EU financial stability, “Fit-For-55” climate stress test shows | European Banking Authority