Good Practice Guide To Climate Stress Testing

UNEP FI’s
Comprehensive
Good Practice Guide
to Climate Stress
Testing
Contents
Acknowledgements................................................................................................................................. 6
Acronyms ............................................................................................................................................... 7
1. Executive summary.....................................................................................................................10
1.1 Overview................................................................................................................................... 10
1.2 Key takeaways .........................................................................................................................11
1.2.1 Team organisation and skills.................................................................................................. 11
1.2.2 Data requirements and collection.........................................................................................14
1.2.3 Scenarios and models.............................................................................................................15
1.2.4 Outputs and applications .......................................................................................................18
2. Introduction: The increasing prevalence and relevance of climate stress testing......................20
2.1 Importance of addressing climate risks................................................................................ 20
2.2 History of stress testing ..........................................................................................................22
2.3 Overview of the UNEP FI TCFD programme and climate stress testing...............................23
3. Current state of climate stress testing........................................................................................25
3.1 Comparison of stress tests.....................................................................................................25
3.1.1 Purpose of traditional and climate stress tests..................................................................25
3.1.2 Comparing key features of traditional and climate
stress tests ...............................................................................................................................26
3.2 Top-down and bottom-up approaches to stress tests...........................................................33
3.3 Climate stress testing regulatory landscape..........................................................................34
3.3.1 List of recent announcements and initiatives ....................................................................34
3.3.2 Summary of selected international climate stress
testing frameworks.................................................................................................................. 37
3.3.3 Importance of internal climate stress tests........................................................................45
4. Team organisation and skills......................................................................................................46
4.1 Team organisation...................................................................................................................47
4.1.1 Institution-wide team engagement for climate
stress testing............................................................................................................................. 47
4.1.2 Teams and their respective roles........................................................................................... 47
4.1.3 Team resourcing.......................................................................................................................53
4.2 Knowledge and skills...............................................................................................................53
4.2.1 In-house capabilities................................................................................................................53
4.2.2 Building skills through external support...............................................................................58
4.3 Institutional organisation and governance.............................................................................59
4.3.1 Changes to team organisation for climate stress testing................................................59
4.3.2 Regulatory guidance on firm-wide integration of climate stress tests..........................62
UNEP Finance Initiative’s Comprehensive Good Practice Guide to Climate Stress Testing 2
Contents
5. Data requirements and collection............................................................................................... 67
5.1 Data requirements for climate stress testing........................................................................ 68
5.2 Good practices for collecting internal and external data for climate stress testing............76
5.2.1 Good practice steps for data collection...............................................................................76
5.2.2 Guidance on data collection for counterparty analysis.....................................................78
5.2.3 Types of third-party data available........................................................................................85
6. Models and scenarios.................................................................................................................90
6.1 Climate scenarios for stress testing...................................................................................... 90
6.1.1 Overview of climate scenarios...............................................................................................90
6.1.2 Regulatory scenarios and the NGFS..................................................................................... 97
6.1.3 Scenario design and expansion...........................................................................................102
6.2 Models for climate stress testing........................................................................................ 109
6.2.1 Types of models used........................................................................................................... 109
6.2.2 Challenges of using existing models for climate
stress testing.......................................................................................................................... 115
7. Outputs and applications of climate stress testing .................................................................. 117
7.1 Expected outputs of regulatory climate
stress tests.............................................................................................................................117
7.2 Applications and uses of climate stress tests outputs for supervisors.............................121
7.3 Recommendations on the use of climate stress tests outputs for financial institutions.126
7.3.1 Overarching recommendations for the climate stress
test process.............................................................................................................................126
7.3.2 Leveraging climate stress tests outputs .......................................................................... 128
8. Conclusion and suggested enhancements ..............................................................................140
8.1 Conclusion ............................................................................................................................ 140
8.2 Next steps for climate stress testing................................................................................... 142
9. Appendices................................................................................................................................144
Appendix 1........................................................................................................................................ 144
Appendix 2........................................................................................................................................ 144
Appendix 3........................................................................................................................................ 145
Appendix 4........................................................................................................................................ 145
Appendix 5........................................................................................................................................ 146
10. Glossary.................................................................................................................................... 147
11. Bibliography..............................................................................................................................150
Contents
Figures and tables
Figure 1: Functions and teams (adapted from ACPR, 2020)........................................................................12
Figure 2: Modelling Approach for Climate Stress Testing............................................................................. 17
Figure 3: Potential financial impacts of physical and transition risks (CICERO, 2017)............................ 21
Figure 4: Functions and teams (adapted from ACPR, 2020)........................................................................48
Figure 5: Recommended roles for the climate risk team..............................................................................60
Figure 6: TCFD’s recommendations for governance......................................................................................63
Figure 7: Bank lending for emission-intensive sectors.
Figure 8: Survey results (2021) showing hardest data types to collect according to
participating institutions in UNEP FI’s TCFD programme............................................................ 71
Figure 9: Key challenges in collecting data for climate stress testing........................................................72
Figure 10: Data granularity of climate-related risks as identified by the NGFS
(NGFS, 2021).........................................................................................................................................75
Figure 11: Examples of low-carbon transition initiatives that clients can participate in........................... 81
Figure 12: Custom emissions factors for companies......................................................................................83
Figure 13: Example of Scope 1 emissions calculations...................................................................................83
Figure 16: Example GHG emissions summary for a company.......................................................................84
Figure 17: Key features of the ECB climate stress............................................................................................86
Figure 18: Classification of the NGFS Reference Scenarios...........................................................................92
Figure 19: Storylines of the Phase I NGFS Reference Scenarios...................................................................92
Figure 20: Comparison of Phase I and Phase II NGFS Scenarios..................................................................93
Figure 21: Summary of impacts of the CBES Scenarios.............................................................................. 100
Figure 22: Decrease in GDP in the late action scenario.................................................................................101
Figure 23: UK primary energy mix for current, early action and late action scenarios............................102
Figure 24: Key Processes for Scenario Analysis in a Climate Stress Test................................................ 103
Figure 25: Key variables indicative list by the Bank of England................................................................... 105
Figure 26: Modelling approach for climate stress testing............................................................................ 110
Figure 27: The modelling framework developed by BdF/ACPR for the pilot exercise............................ 114
Figure 28: A schematic representation of the transition and physical risk scenarios
included in the ACPR exercise........................................................................................................ 119
Figure 29: Evolution of the sectoral structure of credit exposures under the sudden
transition scenario............................................................................................................................. 120
Figure 30: Evolution of the aggregated annual cost of risk for
participating banks............................................................................................................................ 120
Figure 31: Weighted average of the probabilities of default (PD) by sector over time.............................121
Figure 32: Desired outcomes of CBES.............................................................................................................. 125
Figure 33: Anticipated primary uses of the climate stress test results by the
UNEP FI TCFD programme participants........................................................................................127
Figure 34: Six key uses of a climate stress test.............................................................................................. 129
Figure 35: Transmission channels for climate risks into financial risks (NGFS, 2021)............................132
Table 1: Comparison of key features of traditional and climate stress testing....................................... 27

Table 2: Overview of top-down and bottom-up stress testing....................................................................33
Table 3: List of announcements and exercises by countries for climate scenario
and climate stress testing exercises (NGFS, 2021).......................................................................35
Table 4: Comparison of key regulatory climate stress tests....................................................................... 41
Table 5: Suggested involvement of teams and their responsibilities in climate stress testing............49
Table 6: Recommended key climate-specific skills required for teams....................................................54
Table 7: Recommended key climate-specific knowledge areas for teams..............................................55
Contents
Table 8: Types of traditional macro-financial data required for climate stress testing..........................68
Table 9: Types of climate-related data required for climate stress testing..............................................68
Table 10: Climate data required by firms to assess counterparty-specific climate risks........................79
Table 11: Climate data accessible through third-party providers.................................................................85
Table 12: Open data sources available for physical risk hazards.................................................................88
Table 13: Open data sources for emissions data............................................................................................89
Table 14: Comparison of NGFS scenarios to other climate scenarios.......................................................96
Table 15: Recommended variables for climate stress testing....................................................................106
Table 16: Commonly used models for climate scenario analysis............................................................. 110
Exhibit 1: Monetary Authority of Singapore’s Action Plan to build knowledge and

capabilities in sustainable finance....................................................................................................58
Exhibit 2: Bank of England (BoE) and Prudential Regulation Authority.......................................................64
Exhibit 3: Danmarks Nationalbank interlinking of climate and financial data............................................70
Exhibit 4: Availability—A key data issue identified by the NGFS....................................................................73
Exhibit 5: Quality—A key data issue identified by the Basel Committee on
Banking Supervision (BCBS)..............................................................................................................73
Exhibit 6: Reliability—A key data issue identified by the NGFS......................................................................74
Exhibit 7: Comparability—A key data issue identified by the NGFS..............................................................74
Exhibit 8: Granularity—A key data issue identified by Basel Committee on
Banking Supervision (BCBS)..............................................................................................................75
Exhibit 9: Limited internal expertise—A key data issue identified by BCBS................................................76
Exhibit 10: Use of historical data—A key data issue identified by BCBS........................................................76
Exhibit 11: Counterparty data guidance by the Bank of England.................................................................... 81
Exhibit 12: Example—GHG Emission Calculation Tool by the Greenhouse Gas Protocol..........................82
Exhibit 13: Internal data collection by the European Central Bank.................................................................86
Exhibit 14: Bank of England: Drawing Climate Scenarios from NGFS for the Climate
Biennial Exploratory Scenario (CBES).............................................................................................101
Exhibit 16: Qualitative and quantitative outputs from ACPR 2020 climate pilot exercise....................... 119
Exhibit 17: Key takeaways on climate stress testing outcomes from the BIS report
‘Stress-testing banks for Climate Change—a Comparison of Practices’.................................122
Exhibit 18: Desired outcomes of the 2021 CBES by the BoE and the PRA................................................ 124
Exhibit 19: Probing the tails with Reverse Stress Tests................................................................................. 134
Exhibit 20: Adjustment to credit ratings and pricing of climate risks...........................................................137
Contents
Acknowledgements
Authors
UN Environment Programme Finance Initiative (UNEP FI)
David Carlin, TCFD Programme Lead
Maheen Arshad, TCFD Programme
Emily Fraser, Copy Editor
External Collaborators
Vincent Noinville, Independent Expert
Project Management
The project was set up, managed and coordinated by UNEP FI, specifically: Remco
Fischer and David Carlin.
The pilot project was led by a Working Group of 45 banks and investors convened by
UNEP FI:
ABN-AMRO Credit Suisse MUFG
Access Bank Danske Bank NAB
AIB Desjardins NatWest
Bank of America DNB NIB
Bank of Ireland EBRD Rabobank
Banorte FirstRand RBC
Barclays FTF Santander
BBVA ING Scotia Bank
Bentall Green Oak Intesa Sanpaolo Standard Bank
BMO Investa Storebrand
Bradesco Itau TD Asset Management
Caixa Bank KB FG TSKB
CDL KBC UBS
CIB Linkreit Wells Fargo
CIBC Manulife
Citibanamex Mizuho
Acknowledgements
Acronyms
ACPR Autorité de contrôle prudentiel et de résolution (France)

AFM Authority for the Financial Markets (Netherlands)
AI Artificial Intelligence
AIM Asia-Pacific Integrated Model
ALM Asset and Liquidity Management
APRA Australian Prudential Regulation Authority
AR5 Fifth Assessment Report
AR6 Sixth Assessment Report
BCBS Basel Committee on Banking Supervision
BdF Banque de France
BIS Bank for International Settlements
BOC Bank of Canada
BoE Bank of England
BPI Bank Policy Institute
CAIT Climate Analysis Indicator Tool
CARMA Carbon Monitoring for Action
CA100+ Climate Action 100+
CBES Climate Biennial Exploratory Scenario
CBIRC China Bank and Insurance Regulatory Commission
CCS Carbon capture and storage
CDC Centers for Disease Control and Prevention (USA)
CDP Carbon Disclosure Project
CDR Carbon Dioxide Removal
CEO Chief Executive Officer
CFRF Climate Financial Risk Forum
CICERO Centre for International Climate Research (Norway)
CMIP Coupled Model Intercomparison Projects
CO2 Carbon Dioxide
COP Conference of the Parties
COP26 Twenty-sixth session of the Conference of the Parties to the UNFCCC
CVA Climate Vulnerability Assessment
DNB De Nederlandsche Bank
EAD Exposure at Default
Acronyms
EBA European Banking Authority
ECB European Central Bank
ESG Environmental, Social and Governance
EU European Union
FCA Financial Conduct Authority (UK)
FSAN Financial Supervisory Authority of Norway
FSB Financial Stability Board
GAR Green Asset Ratio
GDP Gross Domestic Product
GHG Greenhouse Gas
GFC Global Financial Crisis
GFDRR Global Facility for Disaster Reduction and Recovery
G20 Group of Twenty
HKMA Hong Kong Monetary Authority
HQLA High quality liquid assets
ICAAP Internal Capital Adequacy Assessment Process
IEA International Energy Agency
IFF Institute of International Finance
IFRS International Financial Reporting Standards Foundation
IIASA International Institute for Applied Systems Analysis
IIF Institute of International Finance
ILAAP Internal Liquidity Adequacy Assessment Process
IMF International Monetary Fund
IPCC Intergovernmental Panel on Climate Change
ISIMIP Inter-Sectoral Impact Model Intercomparison Project
IT Information Technology
ITS Implementing Technical Standards
I4CE Institute for Climate Economics
JGCRI Joint Global Change Research Institute
KNMI Koninklijk Nederlands Meteorologisch Instituut (Royal Netherlands
Meteorological Institute)
KPIs Key Performance Indicators
LGD Loss Given Default
MAS Monetary Authority of Singapore
NDCs Nationally Determined Contributions
NGFS Network for Greening the Financial System
NGO Non-Governmental Organisation
NIES National Institute for Environmental Studies
NIESR National Institute of Economic and Social Research
NZ Net Zero
NZAOA Net-Zero Asset Owner Alliance
Acronyms
NZE2050 Net-Zero Emissions 2050
OSFI Office of the Superintendent of Financial Institutions (Canada)
PBL Netherlands Environmental Assessment Agency
PBOC People’s Bank of China
PCAF Partnership for Carbon Accounting Financials
PD Probability of Default
PFH Planning for Hazards
PIK Potsdam Institute for Climate Impact Research
PRA Prudential Regulation Authority
PREP Partnership for Resilience and Preparedness
PRI Principles for Responsible Investment
RBNZ Reserve Bank of New Zealand
RCP Representative Concentration Pathway
RST Reverse Stress Tests
RWA Risk-Weighted Assets
SCAP Supervisory Capital Assessment Program (USA)
SCMP South China Morning Post
SDS Sustainable Development Scenario
SMF Senior Management Function
SMU Singapore Management University
SREP Supervisory Review and Evaluation Process
SSM Single Supervisory Mechanism
SSP Shared socioeconomic pathway
SSRN Social Science Research Network
STEPS Stated Policies Scenario
TCFD Task Force on Climate-Related Financial Disclosures
TVFs Transition Vulnerability Factors
UK United Kingdom
UN United Nations
UNEP UN Environment Programme
UNEP FI UN Environment Programme Finance Initiative
UNFCCC United Nations Framework Convention on Climate Change
USA United States of America
USD United States Dollar
WEM World Energy Model
WEO World Economic Outlook
WIOD World Input-Output Database
WRI World Resources Institute
Acronyms
1. Executive summary
1.1 Overview
Across the world, financial actors have acknowledged the growing risks of climate
change to financial stability. Financial regulators, such as central banks, have a clear
mandate to ensure financial stability, and as such, must be able to systematically
assess climate risks. To understand exposures, risks and resiliency, these supervisors
have turned to a common tool: scenario-based stress testing. After the Global Finan-
cial Crisis (GFC), traditional stress tests became an integral part of a financial insti-
tution’s risk management toolkit with regulators conducting large-scale supervisory
stress tests.
A climate stress test is a forward-looking exercise designed to measure a financial
institution’s exposure to climate risks, using scenario analysis including severe climate
risks, to assess the potential impact of climate change on the institution’s business
model. Climate stress tests may leverage significant elements of traditional capital
stress testing, but also contain a number of important differences. Institutions can and
should leverage the knowledge and skills developed through years of post-GFC stress
testing, but will also need to adapt to meet the emerging challenges of conducting a
climate stress test.
This report, published as part of United Nations Environment Programme Finance
Initiative (UNEP FI) Phase III Task Force on Climate-Related Financial Disclosures
(TCFD) programme, aims to provide a detailed user guide for financial institutions look-
ing to understand the nature of new climate stress tests and develop effective plans for
executing them. The report was created to assist financial institutions in their climate
stress testing journey and should be adapted to meet the needs of a given firm.
Financial practitioners who read this report will benefit from exposure to five major
content areas:
◾ Emerging climate stress testing requirements and the landscape of test structures;
◾ Best practices for climate stress testing team organisation and the skills required;
◾ Best practices for climate stress testing data requirements and collection;
◾ Best practices for selecting the scenarios and models to use in climate stress test-
ing; and
◾ Best practices for how climate stress testing outputs can and should be applied to
meet regulatory requirements and produce useful internal insights.
Executive summary
1.2 Key takeaways
1.2.1 Team organisation and skills
Key messages
◾ Institution-wide team engagement is required for climate stress testing, with
firms assigning responsibilities to a large number of teams from across the
organisation.
◾ Firms should devote appropriate resources (financial and human) to the execu-
tion of climate stress testing, with a specific focus on the activities of data
collection and analysis, model development and strategic planning.
◾ A large body of knowledge and skills will need to be developed in-house to reach
a satisfactory level of proficiency in running climate stress tests.
◾ Climate-related training and knowledge development programmes need
to be geared towards a diverse set of teams across the firm, rather than just
client-facing employees.
◾ Financial institutions will need to make substantial changes to their organisa-
tion including the development of a climate risk team with adequate resources,
executive sponsorship and authority to oversee, coordinate and manage the
◾ Adequate resourcing should be implemented to integrate climate stress testing
into the firm’s organisational structure and processes with robust governance
and oversight.
Climate stress testing is a full institutional endeavor. It demands engagement from

business lines, risk teams and support from senior leadership. This report explores the
different roles and skills those teams need to play in executing a climate stress test.
As shown below, a diversity of teams and skills are involved in both traditional stress
testing and climate stress testing. However, climate stress testing also demands new
skills and new modes of engagement from traditional stress testing.
Executive summary
Climate risk, Stress testing, Business line, Credit
Climate risk, Stress testing, Research, Business
Risk modelling, Model Validation, Credit risk, risk, Market risk, Operational risk, Country risk,
Climate risk, Research, line, Risk modelling, Model Validation, Credit
Market risk, Operational risk, Country risk and Credit risk and market risk portfolio, Asset
Business line teams risk, Market risk, Operational risk and Country
Asset and Liquidity management teams and Liquidity management and Strategy and
risk teams
Planning teams
Scenario design Financial risk assessment Outcome
Counterparty credit risk
Impact on default
rate
Economic or finan- Macroeconomic
cial shocks impacts
Impact on lenders’ Total impacts on
Climate
earnings banks
Climate related Sectoral & infrasec-
shocks toral impacts
Impact on asset
prices
Liquidity risk
Macro feedbacks
Climate feedbacks
Internal audit and Change Standard modelling structure

Function and IT teams Climate change augmented structure
Figure 1: Functions and teams (adapted from ACPR, 2020).
Executive summary
Climate stress testing will require these teams and functions to have varying levels of
proficiency in key climate-specific skills and knowledge to ensure the effective design
and execution of the exercise. Adequate resources must be devoted to the develop-
ment of institutional knowledge around climate risks and climate data.
In order to do so, firms should establish staff roles and responsibilities related to
climate stress testing, develop policies and procedures inclusive of these roles, appoint
climate risk specialists/ambassadors in each team and provide adequate training
by experts. It is also recommended that firms establish a firm-wide programme for
climate knowledge development to further build organisational expertise. Firms may
also rely on external support for skills or knowledge they lack in-house, in the form of
external consultants and contractors, national and industry associations and interna-
tional initiatives.
It is important that climate stress tests are embedded into a firm’s current risk
processes, with senior management and the board actively engaged in the governance,
implementation and management of the programme. It is also recommended that a
firm-wide risk committee be established for climate risk.
Financial institutions will have to make substantial changes to integrate climate risk
assessment into existing risk and stress testing frameworks. Recommended good
practices for embedding climate stress testing into current institutional frameworks
include:
◾ Developing a climate risk team with adequate resources, executive sponsorship and
authority to oversee, coordinate and manage the climate stress test;
◾ Establishing a climate risk research and intelligence function;
◾ Managing climate-related data centrally;
◾ Fostering collaboration among a range of senior experts; and
◾ Establishing a cross-functional project team to design and execute the climate
stress test.
Executive summary
1.2.2 Data requirements and collection
Key messages
◾ Data needed for a climate stress test can be divided into two main cate-
gories—traditional macro-financial (macroeconomic as well as financial)
data and climate-related (industry-specific, weather-related) data. However,
collecting climate data can be a challenge for many financial institutions.
◾ For effective climate stress testing, firms should develop policies and proce-
dures to support the collection, processing and use of climate data.
◾ Good practices for collecting data involve identifying data needs, understand-
ing the availability of data sources, implementing industry standards, validat-
ing data, identifying data gaps and adapting institutional systems.
◾ Best practices for securing climate data on and from clients can be divided
into three categories: (i) steps firms can take in-house and through collabo-
rations, (ii) steps requiring client engagement, and (iii) steps to build robust
data collection processes in the future.
◾ When collecting emissions data, firms should identify internal sources
through which emissions data for clients can be accessed and develop inter-
nal tools to calculate emissions data.
◾ When collecting climate data from external data providers, it is recommended
that firms develop an internal policy, identify easily accessible open-source
platforms, develop a questionnaire based on the institution’s data needs and
actively communicate with data providers.
◾ Adapt existing downstream models, for example loss forecasting models, to
incorporate climate-related variables
Collecting climate-related data is a new demand for many institutions that brings with
it new challenges. Issues may be encountered due to limited data availability, lack of
data quality, limited granularity and coverage, limited comparability and standardisa-
tion and difficulty integrating the data into financial processes. In order to overcome
these obstacles, firms should develop policies and procedures to support the collec-
tion, processing and use of climate data.
Executive summary
Good practice steps for collecting internal and external data include:
◾ Identifying the firm’s data needs for the exercise;
◾ Exploring a wide range of sources and selecting the most suitable;
◾ Implementing industry standards;
◾ Establishing and documenting a clear data validation process;
◾ Identifying data gaps; and
◾ Adapting institutional systems and developing methodologies to implement policies
and procedures for data collection.
Counterparty-level modelling demanded by some climate stress tests poses its own
unique set of challenges. Below we provide recommended guidance for collecting
client data for counterparty analysis.
Steps firms can take ◾ Attempt to collect as much climate data on the client using
in-house and through in-house capabilities.
collaborations ◾ Increase stakeholder collaboration by collaborating with regula-
tors, municipalities and governments.
◾ Develop industry partnerships to provide tools and support to
clients to produce data.
Steps requiring client ◾ Take part in open and effective communication with the client on
engagement the required data.
◾ Update current client engagement processes by integrating data
requirements for climate stress testing.
Steps to build robust data ◾ Develop questionnaires based on data needs to act as a guide
collection processes in the for both the bank and the client for collecting necessary informa-
future tion.
◾ Integrate the use of Geographic Information Systems (GIS) to
gather geospatial data.
Financial institutions may also need to collaborate with external data providers to
access required data. External providers include open data sources, commercial data
sources, scenarios and models. When gathering climate risk data from external data
providers, it is recommended that firms:
◾ Develop internal policies for gathering climate data from external data providers;
◾ Identify potential data providers and open-source platforms to access required data;
◾ Develop a questionnaire based on the institution’s data needs to determine coverage
and methodologies; and
◾ Actively communicate with data providers, including taking part in initiatives
designed to assist data collection.
Executive summary
1.2.3 Scenarios and models
Key messages
The modelling process in a climate stress test consists of modelling climate
variables, measuring the impact of climate risks on macroeconomic variables,
breaking down the macroeconomic impacts per sector or per portfolio and
quantifying the impact on the financial institution. Firms should:
◾ Adapt reference scenarios geographically and sector-wise and derive the
impact on key drivers to develop a relevant scenario narrative;
◾ Implement a process to integrate the latest scenario developments and
emerging methodology standards into the scenario design for climate stress
testing;
◾ Undertake scenario expansion to extrapolate the additional scenario vari-
ables required for the climate stress test; and
◾ Improve modelling capabilities, through identifying gaps in current in-house
models and enhancing communication with external modelers and academ-
ics on scenario expansion and understanding of different models.
Climate scenario analysis is at the core of climate stress testing. Scenario analysis for
a climate stress test includes four main steps—scenario selection, variable selection,
modelling for risk quantification, and using the outputs for risk assessment. Scenario
and variable expansion are often required to produce credible outputs for diverse geog-
raphies, sectors and asset types.
Climate scenarios include physical and transition risk variables which are then
combined with macroeconomic and financial variables to quantify the impact of
climate risks in a specific scenario. Climate risk variables commonly include physical
hazards, carbon price, energy price, energy consumption and greenhouse gas (GHG)
emissions. Macroeconomic variables commonly include gross domestic product
(GDP), unemployment and inflation. Financial institutions may be required to undertake
expansion to extrapolate additional variables for the scenarios. It should be noted that
some current internal and regulatory climate stress tests may look at a single aspect
of climate risk (transition or physical risk). While this may not give a comprehensive
picture of all climate risks faced, these exercises still yield a large amount of valuable
information for a firm.
When using climate scenarios, good practices for selection, expansion and assess-
ment include the following:
1. Conduct portfolio analysis to understand which portfolios need to be stressed for
climate risks;
2. Select a relevant scenario horizon;
3. Identify key climate and economic drivers;
4. Develop criteria for selecting reference climate scenarios;
5. Determine relevant regions and sectors for the scenario;
6. Consider assumptions in relation to the scenario pathways;
Executive summary
7. Adapt/expand the reference scenarios geographically and sector-wise if granular-
ity is insufficient;
8. Derive the impact on the identified key drivers to develop a relevant narrative;
9. Assign responsibility to a team to keep up-to-date with the latest developments in
scenario development; and
10. Implement a process to integrate these developments into the scenario design.
The modelling process in a climate stress test requires the use of a combination of
various models. Figure 2 below provides a break-down of the modelling approach for a
Physical risk variables Transition risk variables

Climate Scenario Model
(e.g., temperature rise, flood (e.g., government policies,
(IAM models)
risk, drought) technology investment)
Macrofinancial variables
Macroeconomic Model
(e.g., inflation, employment)
Sector-level disaggregation
Sectoral and Portfolio Models
Climate risks for segments of a portfolio
Financial Models Loss Given Default (LGD) Probability of Default (PD)

(Rating models and counterparty effects (e.g., reduction in effects (e.g., higher costs,
specific credit models) property value) unemployment)
Probability weighted Potential impact of climate

Expected Credit Loss (ECL) risks on business model (e.g.,
Risk-weighted assets profitability, capital & liquidity)
Figure 2: Modelling Approach for Climate Stress Testing.
When conducting modelling for a climate stress test, it is recommended that firms:
◾ Determine the parameters required for the modelling;
◾ Define the objective, inputs and outputs and the required components for each
model in the approach;
◾ Develop a range of criteria to determine the types of models to be used for the anal-
ysis;
◾ Survey existing models and providers to onboard;
◾ Identify gaps in the current in-house models;
◾ Increase coordination with external modelers and academics for scenario expan-
Executive summary
sion and to improve understanding of different models;
◾ Improve collaboration with local modelers and institutions; and
◾ Further develop in-house modelling capabilities aligned with the guidance and speci-
fications provided by supervisory authorities.
1.2.4 Outputs and applications
Key messages
◾ Supervisory climate stress tests have a wide range of applications, includ-
ing being a learning exercise to mobilise and raise the firm’s awareness of
climate risks, encouraging boards to understand the challenges and take a
strategic approach to managing risks, improving a firm’s climate risk manage-
ment, modelling and data, and improving a firm’s climate-related disclosures.
◾ Firms need to ensure clear, consistent and wide-ranging coordination and
ensure climate stress tests are a key opportunity to learn and trigger action.
◾ A climate stress test approach should initially be kept simple and focused on
material exposures.
◾ Results can be further leveraged to improve a firm’s climate risk manage-
ment and help set the risk appetite, support the firm’s strategy, customer
engagement and investment, and support the firm’s external disclosures and
compliance.
Climate stress testing can produce a wealth of qualitative and quantitative outputs of
interest to both supervisors and internal stakeholders. There are a number of ways in
which climate stress test results can be effectively used to better manage overall finan-
cial system risks, as well as individual institutional climate risks. Current applications
for climate stress tests include:
◾ Mobilising and raising an institution’s awareness of climate-related risks;
◾ Encouraging boards to understand the challenges and take a strategic, long-term
approach to managing climate risks;
◾ Identifying gaps in climate expertise and resources required to assess and manage
climate risks; and
◾ Improving the institution’s climate-related disclosures.
Beyond the immediate benefits of the exercise, climate stress testing outputs
can be leveraged for further key uses, for example to:
◾ Improve understanding of the business’ vulnerabilities, to inform risk appetites and
set targets and limits;
◾ Develop risk management and mitigation strategies;
◾ Inform proposed shifts in business strategy;
◾ Inform customer engagement, especially in vulnerable sectors;
◾ Provide direction for the firm’s climate data and infrastructure investments; and
◾ Provide value-adding management information and enhanced external reporting
and disclosures.
Executive summary
In order to best use the outputs of a climate stress test, climate stress testing exer-
cises need to be effectively communicated across the firm and at all levels of manage-
ment, they should be a learning opportunity and should trigger action. The climate
stress testing approach should be simple and focused on material exposures and the
climate risk assessment should be fully leveraged to drive change and improve busi-
ness performance in light of climate change.
As climate stress testing is still in its early stages, many financial institutions find
conducting a test to be a challenging endeavor. This report provides detailed guidance
on best practices and regulatory expectations for the climate stress testing process.
Going forward, financial institutions, regulators, climate scientists, modelers, research
institutions and other relevant stakeholders should work in harmony to maximise the
potential for climate stress testing to be a tool that promotes financial stability, reduces
climate risks and accelerates the climate transition.
Executive summary
2. Introduction:
The increasing
prevalence and
relevance of climate
stress testing
2.1 Importance of addressing climate risks

As GHG emissions from human activities continue, the Intergovernmental Panel on
Climate Change (IPCC) has warned that the planet has already warmed by over 1˚C
above pre-industrial levels and could warm by nearly 4˚C by the end of century if emis-
sions are not reduced (IPCC, 2021). According to the latest science synthesised in
IPCC reports, continued temperature rise could lead to severe economic, social and
environmental consequences. Recognising the scale of these risks, in December
2015 nearly 200 governments took the initiative to strengthen the world’s response to
climate change by signing the Paris Climate Change Agreement. However, to reach the
emission reductions goals that science says are necessary to keep planetary tempera-
tures within safe levels, fundamental shifts are required across the global economy.
These shifts to a low-carbon future bring with them distinct challenges and risks for
economic sectors.
Financial actors are confronting the looming physical risks of climate change and the
increasing transition risks inherent to a low-carbon transition (Figure 3). Being able to
better price these risks would help promote financial stability, ensure appropriate allo-
cation of capital to support resilience and adaptation, and hasten the transition to a
sustainable future. With these objectives in mind, the Financial Stability Board of the
G20 launched the Task Force on Climate-Related Financial Disclosures (TCFD) to
improve climate-related risk disclosures. The TCFD issued its first recommendations in
2017 and has issued further guidance in the years that have followed.
Introduction: The increasing prevalence and relevance of climate stress testing
Figure 3: Potential financial impacts of physical and transition risks (CICERO, 2017).
Physical risks are risks associated with the impact of changes in weather and climate,
on the economy. Physical risks can be acute in nature, like floods, heatwaves and wild-
fires, or chronic in nature, such as sea level rise, temperature changes and precipitation
changes. The accumulation of acute and chronic physical risks has the potential to
cause damage to physical assets, and also negatively affect sectors that are reliant
on natural resources and disrupt supply chains. Both acute and chronic physical risks
can pose financial risk for firms. For example, physical risk drivers have the potential
to negatively impact a borrower’s ability to repay a loan due to negative wealth effects.
Climate events can cause damage to physical assets of banks’ counterparties caus-
ing a reduction in the asset value. Damaged assets will lead to a reduction in income
which increases the likelihood of a borrower defaulting, posing credit risk for the firm
(BIS, 2021). Weather events increasing in frequency, severity and uncertainty also have
the potential to create high volatility in financial markets resulting in market risks, and
have the potential to disrupt operational ability increasing costs for firms (BIS, 2021).
Transition risks are related to changes in legislation, policies, technology and the
market, during the transition towards a low-carbon economy (TCFD, 2017). For
example, transition risks can be related to a growing number of jurisdictions that are
implementing policies to reduce emissions, technological innovation for less carbon-in-
tensive technology, increasing consideration of climate risks by investors in their deci-
sion-making and shifts in consumer sentiments towards lower carbon emissions
(BIS, 2021). If a business model of a firm does not adapt to a low-carbon economy,
it has the potential to face losses (IMF, 2019). More explicitly, shifts in asset values
and higher costs may lead to lower profit margins (IMF, 2019). Transition risks can also
give rise to financial risks for financial institutions. For example, implementation of a
carbon tax can increase operating costs for businesses, reducing earnings which can
impact their ability to repay loans. Similarly, a low-carbon transition can lead carbon-in-
tensive assets, such as fossil fuel reserves and coal mines, to become stranded assets
and devalue, impacting business income. Changes in policies, technological advances
and investor behavior can impact borrowing costs and cause a sudden repricing of
financial assets. A rapid transition could also lead to sudden shifts in prices, resulting
in market risks for firms (BIS, 2021). A transition to a low carbon economy also has the
potential to impact countries which receive most of their income from fossil fuels.
Significant losses in GDP, vulnerable infrastructure, lowered productivity and the vulner-
ability of carbon-reliant sectors, are just a handful of ways in which climate risks will
impact the global economy. In addition to economic impacts, it is also important not
to underestimate the health and social impacts of climate change. Climate change
can limit food and water supply access, causing malnutrition and diarrheal disease.
Environmental and living degradation caused by climate change can also lead to inter-
nal struggles—such as civil conflicts, forced migrations and an overall perpetuation
of social inequality. Further, increased pollution may worsen cardiovascular diseases
and respiratory complications, extreme heat may give rise to heat-related illnesses,
and severe weather events can lead to injuries, fatalities and negative mental health
impacts (CDC, 2021).
Understanding the financial implications of climate change, central banks and poli-
cymakers have begun working towards gaining insights into the potential impacts of
climate risks on the financial system. One way to do this is by developing stress-testing
methodologies that identify climate risks. This report provides best practice guidance for
financial institutions on conducting a climate stress test. Section 3 of the report gives
an overview of the regulatory climate stress testing landscape and how climate stress
tests differ from traditional stress tests. The remaining sections present recommended
guidance on four key components for conducting a climate stress test—(1) team organ-
isation and skills, (2) data requirements and collection, (3) scenarios and models and (4)
applications of the outputs/results. Together these sections provide clear and concise
recommendations on best practices for conducting a climate stress test.
2.2 History of stress testing

Scenario-based stress testing was first adopted by financial institutions in the 1990s
but was only used in small-scale isolated exercises with risk managers using histori-
cal events as worst-case scenarios. In 2008, the GFC resulted in severe stress caus-
ing numerous metrics to reach levels that had not been witnessed since the Great
Depression, with regulators realising that the failure of financial institutions could lead
to widespread economic harm. The GFC helped identify key gaps in risk management
practices across the financial industry at the time, highlighting the need to develop
stress testing frameworks with scenarios of severe stress.
Following the recession, stress testing transitioned to a large-scale, comprehensive
programme overlooked by regulatory authorities. The first example of a large-scale
exercise was the US Supervisory Capital Assessment Program (SCAP) in 2009, which
has been recognised as contributing to the stabilisation of the US financial system.
The SCAP was later followed by traditional stress test exercises in other countries,
including the European Banking Authority (EBA) and the Bank of England (BoE) (Bank
of England, 2016).
Now, traditional stress testing is an integral part of a financial institution’s toolkit. A
traditional stress test is used to measure the capital resiliency of a financial institution
to severe, hypothetical scenarios. The results of a stress test are then used by central
banks and regulators to understand risks, adjust capital requirements and develop
policy for financial resilience. Over recent years, the stress test has been developed to
measure a wide range of resilience metrics (Bank of England, 2016).
Appreciating that stress testing has become a vital tool for risk managers to under-
stand if an institution can withstand severe situations, firms have started using stress
testing as a methodology for assessing climate risks. Since assessing climate risks
requires forward-looking data and stress tests are forward-looking exercises, they are
well designed to measure a firm’s exposure to climate risks and their potential impact
on business strategies (BIS, 2021), contrary to traditional risk management techniques
that rely on historical data and statistical risk modelling.
The use of stress tests to measure the potential impact of climate risks on the finan-
cial sector is rapidly evolving (Deloitte, 2020). With the growing uptake of the recom-
mendations by the TCFD, financial institutions have been adapting methodologies to
more specifically quantify their exposure to climate risks.’ The first climate stress test-
ing exercises by central banks and supervisors took place under the initiative of the
Network for Greening the Financial System (NGFS), with the aim of raising awareness
of climate risks among firms and to expand methodologies for measuring climate
risks. Though climate stress testing is still in its early stages, several jurisdictions have
now announced their intention to conduct climate stress tests (see section 3.3.1)
(I4CE, 2021).
2.3 Overview of the UNEP FI TCFD programme

and climate stress testing
Since the publication of the Financial Stability Board’s (FSB) TCFD recommendations
in 2017, UNEP FI has convened pilot programmes with a consortium of banks and
investors to assist them in implementing the TCFD framework and issuing meaning-
ful climate disclosures. Almost 100 financial institutions (banks, investors and insur-
ers) globally have participated in these pilots and have been supported by nearly a
dozen technical partners from climate modelers to climate risk experts. The pilot
programmes have created numerous tools, frameworks and guides to empower both
participating institutions and those throughout the financial industry to better manage
and disclose their climate risks.
The exercise began with a year-long programme known as ‘Phase I’ of UNEP FI’s TCFD
Programme, involving 16 international banks. The consortium of global banks collabo-
rated to develop assessment approaches for physical and transition risks and oppor-
tunities. The pilot created methodologies that were adaptable and flexible to banks
across geographies and that would promote consistency and comparability.
Following Phase I, Phase II expanded to include 39 banks, to enhance their climate
risk toolkits and improve their climate-related disclosures. The programme worked to
develop a variety of tools, frameworks and thought papers to drive the financial sector
forward in identifying, assessing, managing and disclosing climate risks.
Phase III of the TCFD Programme further expanded to include nearly 50 global banks
and investors. A larger group of participants has provided a range of perspectives
from the financial sector which has led the programme to develop good practices for
climate risk assessment and disclosure.
As part of one of the targeted modules, UNEP FI and industry experts delivered a series
of seminars and interactive discussions on climate stress testing to participants in
the TCFC banking pilot programme. This report aims to synthesise lessons from these
sessions into a guide on best practices for climate stress testing for financial sector
stakeholders.
3. Current state of
climate stress testing
3.1 Comparison of stress tests
3.1.1 Purpose of traditional and climate stress tests

Purposes of a traditional stress test
A traditional stress test helps financial firms forecast their capital positions and infra-
structure robustness under different hypothetical, severe economic shocks. It is
designed to determine whether a bank would have enough financial resources to
withstand extreme economic scenarios in the future and support the economy during
economic downturns. As a result, a traditional stress test can serve various purposes.
Firstly, a traditional stress test allows firms to identify and assess potential risks which
can then be aligned with business strategies and risk management processes at the
portfolio-level and institution-level (BIS, 2018). The exercise can help firms identify risks
which would have otherwise been overlooked. Traditional stress tests can complement
other risk quantification tools, support risk quantification methodologies and provide
perspectives on the validity of statistical models (OSFI, 2009). Traditional stress testing
also plays an important role in supporting capital management for firms by identifying
severe events or changes in market conditions that could impact the institution and
assess its solvency (OSFI, 2009). Similarly, the results can also be used in identifying,
measuring and controlling funding liquidity risks (OSFI, 2009). Results also help regu-
lators identify which firms do not have adequate capital and help prevent them from
defaulting. Authorities also use the results for policy decisions, such as setting mini-
mum capital requirements and bank capital buffer levels (Bank of England, 2016).
Current state of climate stress testing
Purposes of a climate stress test
A climate stress test assesses banks’ vulnerability to the effects of climate change, to
better understand the financial risks that the global financial system faces from global
warming and how banks’ business models could be affected. A climate stress test can
be conducted for various purposes. Most importantly, climate stress testing results
can be incorporated into an institution’s risk management and risk appetite processes.
The test identifies new types of risks over a new risk horizon. Exercises are designed
to identify and address existing gaps in climate risk assessment, including data quality,
and increase awareness among firms of climate risk management (BIS, 2021). Results
can also be used to improve the allocation of assets and assess business strategies
and planning.
3.1.2 Comparing key features of traditional and climate

stress tests
In order to perform forward-looking analysis to assess the vulnerability of financial
institutions to climate-related shocks, key features of a traditional stress test have had
to be adapted. Differences in features between the two test types include scenario
design, time horizon, data and models required, risk types assessed, as well as the
outputs and applications of the exercise. Below we provide a detailed comparison
between a traditional stress test and a climate stress test (Table 1).
Table 1: Comparison of key features of traditional and climate stress testing.
Characteristic Traditional Stress Test Climate Stress Test

Planning ◾ Short time horizon. ◾ Long time horizon.
horizon ◾ Varies between institutions but usually ranges from three to five ◾ Varies between institutions but commonly between 30 and 50
years. years.
◾ Examples: ◾ Longer time horizons let climate risks materialise but increase
◽ US Federal Reserve: Nine quarters. uncertainty.
◽ Bank of England (BoE): Five years. ◾ Integration of short and medium-term effects into the exercise
is important because:
◾ Liquidity stress tests are short, ranging from a day to a few
months. ◽ A low-carbon transition can occur sooner than expected; and
◽ Impacts of physical risks have already begun to materialise.
Balance sheet ◾ Assume bank balance sheet remains static. ◾ Balance sheets can be static or dynamic.
behaviour over ◾ Some assume that the bank balance sheet remains static,
stress horizon however this assumption is unrealistic due to the nature of the
long horizon.
◾ However, strong assumptions are required to model the balance
sheet in the dynamic model.
Number, ◾ Typically, three scenarios: ◾ Typically, between three to five scenarios, as several climate-re-
granularity ◽ Baseline scenario; lated drivers need to be considered.
and types of ◽ Adverse scenario; and ◾ Scenarios include climate change variables (physical and transi-
scenarios used ◽ Can also include a severely adverse scenario.
tion risks).
◾ Possible shocks include change in policy, energy and food
◾ Adverse scenarios are more negative projections than the
prices, technologies, energy demand, or market confidence.
expectations of economic activity and financial market develop-
ments. ◾ Some financial shocks and variables used in traditional stress
tests remain relevant, for example:
◾ One common scenario is the global market shock for financial
institutions and significant trading exposures. ◽ GDP;
◾ Scenarios can mimic the 2008 GFC. ◽ Asset prices; and
◾ Includes macroeconomic and market scenarios (both domestic ◽ Default probabilities.
and foreign): ◾ Need to model interactions between climate, macroeconomy
◽ Unemployment rate; and the financial sector.
◽ Asset prices; ◾ Climate scenarios need to be more granular to effectively anal-
yse borrower-level risks.
◽ Gross Domestic Product (GDP); and
◽ Bond yields.
Data and ◾ Scenarios are calibrated on past events. ◾ Lack of historical data available on the relationship between
models ◾ Historical data available on losses and revenues. climate risk and credit loss.
◾ Loan portfolios analysis includes modelling of: ◾ Scenarios are built on existing research.
◽ Probability of Default (PD); ◾ Speculative judgement is needed on the impact of climate risks
on expected loss.
◽ Loss Given Default (LGD); and
◾ Lots of information and assumptions needed for counterparty
◽ Exposure at Default (EAD).
projections across a long time horizon.
◾ Trading losses:
◾ Models can assess borrower-level risk:
◽ Applies risk-factor shocks to exposures.
◽ Reliable approaches focus on loan losses using PD frame-
◾ Operational risk: work.
◽ Models that relate to losses due to operational risks of ◾ Models have potential to double count the impact of climate
economic conditions. change on asset prices and credit losses.
◾ Pre-provision net revenue modelling. ◾ Results are dependent on emissions, technologies and policy
assumptions for a given sector.
◾ Continued climate stress testing can improve data availability
and help establish the relationship between climate risk factors
& financial information.
Future ◾ A bank must preserve or build up capital reserves. ◾ Currently, authorities publish results in aggregate, therefore
consequences ◾ Banks need to reduce the capital distributed as dividends and individual institutions are not named.
share buybacks to shareholders. ◾ In the future, institutions may be penalised for not managing
◾ Imposition of fines. climate risks.
◾ Non-compliance may raise legal issues & damage firm credibil-
ity.
Impacts of ◾ Intervention by regulators. ◾ Climate stress tests are deemed exploratory at this time.
stress testing, ◾ Public information. ◾ Public information.
including ◾ Depositors can avoid weak banks. ◾ Investors can avoid institutions with high implications of climate
regulatory ◾ Use as a tool for measuring and managing future risks. risks.
effects ◾ To be used to inform the supervisory policy.
◾ Banks model capital effects of tail risks in loan books.
◾ Policy development for a resilient banking system. ◾ Risk management:
◽ Used as a tool for identifying future climate risks;
◽ Banks can manage climate risks in their portfolios and loan
books;
◽ Results can be incorporated into business strategy and plan-
ning; and
◽ Policy can be developed to address and build resilience to
climate risks.
Limitations ◾ Historical data that may originate from aggregate sources with ◾ Lack of historical data, resulting in many assumptions based on
less-detailed data. current research.
◾ Standardised data may skew data results that may be of limited ◾ Potentially faulty data can give rise to false positives or nega-
use at the individual firm-level. tives that can result in under-stressed or over-stressed guide-
◾ Historical data may not fully capture future changes in the lines.
economic environment that can induce shocks.
Risk types ◾ Gauges investment risk in order to measure: ◾ Measures potential impacts of:
◽ Asset viability; ◽ Physical risk; and
◽ Internal processes & controls; ◽ Transition risk.
◽ Portfolio risk; and ◾ Assesses physical & transition risks as risk drivers for financial
◽ Overall institutional resilience. risks.
◾ Assesses the following risks: ◾ Currently, most climate stress tests are designed to assess
credit risk, however there is potential for the test to look at other
◽ Credit risk analysis for loan profiles;
financial risks, such as:
◽ Market risk to assess the potential impact of severe market
◽ Market risk;
events; and
◽ Operational risk; and
◽ Liquidity risk, as stress tests may help gauge which assets
are more likely to inflict losses & damage business opera- ◽ Liquidity risk.
tions. ◾ Additionally, stress tests may also include liability risks, which
◾ Risk governance: may arise due to inadequate action towards climate-related risks.
◽ Results help assess systemic risk potential for the economy.
◾ Results can be used to complement other risk quantification
tools to:
◽ help complement risk quantification methodologies; and
◽ provide perspectives on validity of the statistical models.
Integration of ◾ Asset Management: Firms are allowed to use results to under- ◾ Re-evaluation of business model, internal strategy and planning.
results stand how well assets stand up to market scenarios & external ◾ Risk management:
events, such as:
◽ Insight as to which sectors are vulnerable to physical & tran-
◽ Housing price indexes; sition risks;
◽ Commercial real estate prices; ◽ Re-evaluation of assets held in climate-sensitive sectors &
◽ Equity prices; and classes; and
◽ Stock market volatility. ◽ Improved portfolio metrics going forward.
◾ Risk management: Results can improve business insights by ◾ Credit & price adjustments:
aligning potential risks & losses with proper risk strategies to ◽ Improved insight towards credit rating adjustments; and
minimise loss, for example:
◽ Enhanced ability to predict & model future price of climate
◽ Re-evaluation of sectoral lending; risks.
◽ Enhancement of credit rating systems; and ◾ Sensible lending:
◽ Re-evaluation of risk appetite. ◽ Improved insight towards conditions to include when lending
◾ Firms can use results to gauge investment risk, viability of to vulnerable sectors.
assets & overall investment management.
◾ Results can help establish relationships between different capi-
tal holdings & their sensitivity to economic shocks.
◾ Results may help model customer behavior under extreme
economic events.
◾ Results can help set in place strategies necessary to mitigate
potential losses.
Types of results ◾ Estimated losses & expenses resulting from various types of ◾ Identification of sectors and regions that are exposed to the
provided risks. potential impacts of climate risks.
◾ Effect of revenues & expenditures on income. ◾ Distribution of risks by industries.
◾ Potential effect of future operating, investing & cash position on ◾ Default probabilities to assess credit risk.
cash-flows.
◾ Overall effect of assets, liabilities & capital level on the balance
sheet.
3.2 Top-down and bottom-up approaches to
stress tests
There are two approaches for performing a stress test—a top-down and a bottom-up
approach. A top-down approach is when a supervisory authority performs the test
themselves, using their own framework, which includes a homogenised methodol-
ogy, assumptions, scenarios and models. A bottom-up approach is when a firm uses
its own framework as part of a system-wide or supervisory exercise (BIS, 2018). The
type of approach used can influence the outputs of the stress test exercise. Both
approaches have their benefits and drawbacks. Table 2 below provides an in-depth
overview of both approaches for climate stress testing.
Table 2: Overview of top-down and bottom-up stress testing.
Types of Stress Tests

Top-Down Bottom-Up
◾ Conducted by regulatory authorities or central ◾ Conducted by the firms themselves.
banks. ◾ Firms estimate their exposure to potential
◾ The supervisory authority can define the exer- climate risks.
cise and estimate the magnitude of the impact ◾ It is possible for a regulator to provide guid-
from a climate shock. ance and direction on the scenarios to use,
◾ Participating firms may be asked to map the with the institution then running the scenario
effects of the shock on their assets. analysis and translating them for their coun-
(Climate Risk Review, 2020) terparties.
(Climate Risk Review, 2020; FSB,2020)
◾ Not as resource-intensive as a bottom-up ◾ Highly resource-intensive, requiring firms to
approach for participating financial institutions use models and collect data at the firm-level
(Climate Risk Review, 2020). (Climate Risk Review, 2020).
◾ Premise and constraints of the test are based ◾ Where appropriate, the premise and
on aggregate, macroeconomic assumptions & constraints of the test can be based on a
climate scenarios are adapted to be applicable firm’s own assumptions about what shocks
towards domestic firms (Climate Risk Review, may affect them & their business model
2020). (Climate Risk Review, 2020).
◾ Data is obtained from aggregate sources that ◾ Data originates from own firm and possibly
are generally less granular to cover a wide third-parties, resulting in the use of more
range of participants (Climate Risk Review, granular data for analysis (Open Risk Manuel,
2020). 2016).
◾ Physical risk data is usually collected by geog-
raphy by either using country databases to
determine their vulnerability to physical risks
or using an authority’s own estimates for their
jurisdiction (FSB, 2020).
◾ Transition risk data is collected using official
sector datasets and survey data from financial
institutions (FSB, 2020).
Types of Stress Tests
Top-Down Bottom-Up
◾ Methodologies of a top-down approach can ◾ Firms may have to use their own models to
result in difficulties linking sectors and firms, estimate their exposure to climate risks as a
to the degree to which the action of one sector financial variable, which may require institu-
or firm might affect the risks faced by others tions to develop internal quantitative models
(FSB, 2020). (FSB, 2020).
◾ Disparity in the availability of resources and
capital, as well access to modelling systems,
between financial institutions can result in a
skewed assessment of the financial system as
a whole (Climate Risk Review, 2020).
◾ Test results can be compared firm-to-firm ◾ Test results cannot be compared across firms
due to standardised nature of methodologies since test methodologies can differ between
(Open Risk Manuel, 2016). firms (Open Risk Manuel, 2016).
◾ Individual firms may be less convinced of ◾ Better at capturing firm-specific dynamics due
top-down test results & therefore less likely to to firms’ own estimates of exposure which
implement recommendations based on the increases confidence in the usefulness of the
outputs of the exercise (Climate Risk Review, results and its implementation in addressing
2020). climate risks by firms.
◾ Generalised results present difficulties in ◾ Provides individual firms with the opportunity
linking relevant climate risks to specific firms, to underestimate climate risks in order to
resulting in difficulties in taking appropriate avoid undesirable outcomes.
action by institutions (Climate Risk Review, (Climate Risk Review, 2020)
2020).
◾ Results may give rise to static estimates that
may give limited insight into how risks may
change in the future and therefore can result
in risks being under or over-estimated (FSB,
2020).
Examples: DNB’s energy transition risk stress Examples: BoE’s CBES, and ACPR’s climate pilot
test, and ECB’s economy wide climate stress test. exercise.
3.3 Climate stress testing regulatory landscape
3.3.1 List of recent announcements and initiatives

Globally, a growing number of supervisory authorities have either conducted, are in the
process of conducting or have announced plans to conduct a climate stress test. Table
3 below provides an overview of climate stress testing announcements and initiatives
by the country members of the NGFS.
Table 3: List of announcements and exercises by countries for climate scenario and climate stress testing exercises (NGFS, 2021).
Expected end Balance

Level of Risk
Member date of the sheet Approach Time horizon
granularity coverage
exercise assumption
Autorité de contrôle prudentiel et de Concluded (May Physical,
Hybrid Bottom-up Sector 30 years
résolution (ACPR)/Banque de France 2021) transition
Australian Prudential Regulation Authority Bottom-up, Counterparty, Physical,
Early 2022 Static, hybrid 30 years
(APRA) Top-down macroeconomic, sector transition
Micro-founded
Banca d'Italia Concluded N/A Sector Transition 0 year
approach
Bottom-up,
Banco Central de Chile Q2 2022 Static, dynamic Macroeconomic, sector Transition 5 years
Top-down
Banco de España Dec-21 Static Top-down Macroeconomic, sector Transition 3 years
Top-down, Physical, 30 years, 80 years for GDP
Banco de la República (Colombia) Dec-21 Static Macroeconomic, sector
other transition effects
Counterparty, Physical, 3 years / 20–30 years
Banco de México Dec-21 Static, dynamic Top-down
macroeconomic, sector transition (tbd)
Bangko Sentral ng Pilipinas (Philippines) Mid-2022 Static Bottom-up TBD TBD TBD
Physical,
Bank Al-Maghrib (Morocco) Planning phase Dynamic Other Macroeconomic, sector 30 years
transition
Bottom-up, Counterparty,
Bank of Canada Autumn 2021 Static Transition 30 years
Top-down macroeconomic, sector
May 2022 (sooner
if the bank decides Physical,
Counterparty, "30 years for transition
Bank of England (UK) not to ask for a Static Bottom-up transition,
macroeconomic, sector 60 years for physical"
second round of litigation
submissions)
Physical,
Bank of Korea Dec-22 Static Top-down Sector 30 years
transition
Macroeconomic, sector, Physical,
Bundesbank (Germany) First part: Nov-21 Hybrid Top-down 5–30 years
entity-level transition
Physical, "1 year for flooding risks
De Nederlandsche Bank (Netherlands) Q4 2021 Static Top-down Counterparty
transition 10 years for transition"
Concluded (May Physical,
European Banking Authority Static Top-down Counterparty 30 years
2021) transition
Concluded Physical,
European Central Bank Static Top-down Counterparty 30 years
(September 2021) transition
Physical,
Hong Kong Monetary Authority Dec-21 Static Bottom-up Counterparty, sector 5–30 years
transition
Japan Financial Services Agency/Bank of Counterparty, Physical, "30 years for transition
June-22 Static Bottom-up
Japan macroeconomic, sector transition 80 years for physical"
Malta Financial Services Authority Q2 2022 Static Top-down Sector Transition Short-term horizon
Counterparty, Physical,
Monetary Authority of Singapore H2 2022 Static Bottom-up 30 years
macroeconomic, sector transition
Oesterreichische Nationalbank (Austria) Autumn 2021 Static Top-down Sector Transition 5 years
Bottom-up, "10 years,
People's Bank of China H1 2022 Static Counterparty, sector Transition
Top-down 40 years for macro"
Counterparty, Physical,
Reserve Bank of New Zealand Late 2023 TBD Other 30 years
macroeconomic, sector transition
"Seðlabanki Íslands Physical,
Dec-21 Static Top-down Macroeconomic, sector Not yet decided
(Central Bank of Iceland)" transition
November 2021 for
the current exercise,
South African Reserve Bank Dynamic Bottom-up Sector Physical 3 years
2022-3 for a future
excerices
Suomen Pankki (Bank of Finland) End-2021 Static Top-down Sector Transition 5 years
Superintendencia Financiera de Physical, "10 years for transition
Oct-2021 Static Top-down Sector
Colombia transition 60 years for physical"
Sveriges Riksbank (Sweden) The exercise is in planning phase and details are not determined yet
"First part: end Counterparty,
Swiss National Bank September 2021 Static Other macroeconomic, Transition 5–30 years
Rest: TBD" sector
Blue indicates “concluded”, green indicates “in progress” and grey indicates “in planning”.
3.3.2 Summary of selected international climate stress
testing frameworks
PRA/BoE—United Kingdom
The Climate Biennial Exploratory Scenario (CBES) exercise began in 2021 with results
expected to be published in May 2022 if a second round of the exercise takes place.
Participants include large UK banking groups, life insurers and general insurers,
which account for 70% of the country’s bank lending to households and businesses
and approximately 65% of the UK life insurance market by asset size. The desired
outcomes for the exercise are to (1) size the financial exposures of participants and
the financial system to climate risks, (2) understand the challenges posed to partic-
ipating institutions’ business models, and (3) assist institutions in enhancing their
climate risk management. The CBES scenarios build upon the NGFS climate scenarios
for early policy action, late policy action and no policy action. All three scenarios will
explore both transition and physical risks to a certain degree over the period 2021–
2050. Further, they will measure the impact of the scenarios on their static end‐2020
balance sheets (Bank of England, 2021).
ACPR/BdF—France
The authorities conducted their climate pilot exercise from July 2020 to April 2021.
The pilot exercise was voluntary and carried out by banks and insurers. The exercise
relied on the scenarios by the NGFS and included a baseline scenario corresponding to
an orderly transition and two disorderly transition scenarios. Each of these scenarios
combined different assumptions in terms of the trajectory of the carbon tax and total
productivity levels of factors. Risks were assessed over a 30-year timeframe, covering
the 2020-2050 period. The methodology used static and dynamic balance sheets and
covered both physical and transition risks. The climate stress test was designed to be
a bottom-up exercise with an international dimension and encompassed 55 sectors.
The pilot achieved its objectives of mobilising French banks and insurers, raising aware-
ness about climate risks, quantifying and assessing complex transition or physical risk
scenarios, drawing on the work by the NGFS and providing the first measurement of
risks and vulnerabilities to which French financial institutions are exposed. Results of
the exercise revealed an overall “moderate” exposure of French banks and insurers to
climate risk (ACPR, 2021; ACPR, 2020).
AFM/DNB—Netherlands
DNB conducted a climate stress test in 2018, considering the potential impact of
energy transition risks to the Dutch financial sector. DNB analysed four scenarios: (1)
the policy shock scenario, (2) the technology shock scenario, (3) the double shock
scenario, and (4) the confidence shock scenario (of consumers and investors). The
scenarios were defined to materialise within five years, thus ensuring that the stress
test results are relevant to financial institutions, decision-makers and other stakehold-
ers in the near-term. Furthermore, the stress test only considered transition risks and
not physical risks. The impact of each scenario on Dutch financial institutions was
calculated using data of slightly more than half of the total aggregate exposures of
Dutch banks, insurers and pension funds, from which they concluded that losses were
sizable but manageable (De Nederlandsche Bank, 2018). DNB is currently conducting a
top-down exercise assessing both physical and transition risks with counterparty-level
analysis using a static balance sheet. The time horizon for the exercise is one year for
flooding risk and 10 years for transition risk. The exercise is expected to be completed
in Q4 of 2021 (NGFS, 2021).
FSAN—Norway
In 2019, FSAN conducted a climate stress test on transition risks and published the
results in their Risk Outlook Report 2019. Work on possible transition shocks on the
Norwegian financial system continued in 2020 by the IMF. The IMF estimated the
direct impact of a severe increase in domestic carbon prices on firms under severe
assumptions. Secondly, they mapped the impact of an increase in global carbon prices
on the Norwegian economy via the oil sector. Thirdly, the IMF modelled the impact of a
forced reduction in Norwegian oil firms’ output on shareholder portfolios. Results from
the exercise showed that such a sharp increase in carbon prices would have a signifi-
cant but manageable impact on banks (International Monetary Fund, 2020).
Danmarks Nationalbank—Denmark
Danmarks Nationalbank performed a scenario analysis exercise in 2020 to highlight
transition risks in the banking sector, based on the scenarios by the NGFS. As a precur-
sor to a fully developed climate stress test, DNB conducted sensitivity analyses to
assess whether the banking sector would have a capital shortfall if it faced losses over
a given timeframe. The exercise linked corporations’ accounting data, industry-level
emissions data and credit register data to identify climate risks for corporate lending by
banks. Data for energy labels were also included in the assessment of mortgage lend-
ing. Results indicated that firms were well equipped to handle transition risks. However,
a drastic transition in which the banks need to make large impairment charges over a
short timeframe may result in a capital shortfall (Danmarks Nationalbank, 2020).
ECB—European Union
In March 2021, the ECB released preliminary findings from its economy-wide climate
stress test, designed to help assess the climate risks faced by the financial sector
over the next 30 years. In September, the central bank published an in-depth paper on
the results and methodologies of the exercise. The ECB climate stress test combined
company-level data with exposure data, with the aggregate trajectories for transition
and physical risk embedded into scenarios created by the NGFS. The ECB included
three scenarios (1) orderly transition with limited physical risk, (2) disorderly transi-
tion with limited physical risk, and (3) hot house world with extreme physical risk. The
exercise was conducted entirely by ECB staff and relied upon internal datasets and
models. The exercise highlighted sectors and regions in Europe that may be vulnera-
ble to climate risks. For example, preliminary findings suggest that Southern European
countries are more susceptible to heat stress and wildfires, while Northern European
countries are more vulnerable to flooding. Additionally, mining, energy and manufac-
turing companies are in carbon-intensive sectors and are therefore the most vulnera-
ble to climate-related policy changes. The ECB has also released its methodology for
conducting a bottom-up supervisory climate stress test in 2022 to address institutions’
risks towards climate change and their readiness (ECB, 2021). The results from this
exercise will also be used to inform the climate stress test of the Eurosystem balance
sheet, which is expected to be completed in early 2022 (ECB, 2021; ECB, 2021).
OSFI/BOC—Canada
In November 2020, the BOC and OSFI launched a joint pilot project on climate risk
scenarios to better understand potential transition risks. A report is expected to be
published at the end of 2021 to detail specific scenarios, methodologies, assump-
tions and sensitivities. The project aims further to (1) improve climate scenario anal-
ysis capabilities of institutions, (2) increase the understanding of the financial sector’s
potential exposure to risks associated with transitioning to a low-carbon economy, and
(3) improve the understanding of risk-management practices surrounding climate-re-
lated risks and opportunities. In 2020, four transition scenarios were studied: (1) busi-
ness as usual, (2) nationally determined contributions (NDCs), (3) 2˚C consistent and
(4) 2˚C delayed action. The time horizon selected was 2020 to 2050 (at five-year inter-
vals) (Bank of Canada, 2020). The NGFS progress report on global supervisory and
central bank climate scenario exercises, released in October 2021, provided further
updates on the climate scenario exercise being undertaken by the BOC. The BOC has
assessed market risk using a top-down approach at the sectoral level and used both
a top-down and bottom-up approach for assessing credit risk. The central bank devel-
oped their own scenarios which align with the NGFS scenarios, including a net zero by
2050 scenario (NGFS, 2021).
CBIRC/PBOC—China
The PBOC has announced that it has been conducting climate-related stress tests with
commercial banks, reviewing the implications that climate-related risks may have on
their assets (SCMP, 2021). Key features include a bottom-up and top-down approach
with counterparty-level analysis. The exercise currently being undertaken by the PBOC
includes sector-level granularity, assessing transition risk using a static balance sheet
assumption. The selected time horizon spans from 10 to 40 years (NGFS, 2021).
HKMA—Hong Kong
On 4 December 2020, the HKMA shared guidelines for its pilot exercise on climate
stress tests with participating financial institutions. The pilot exercise is set to take
place in 2021 and aims to assess the banking sector’s resilience to climate change
and facilitate capacity building of banks vulnerable to climate risks. Through the exer-
cise, the regulator hopes that banks will understand the range of climate-related risks,
identify data gaps for risk identification and ultimately establish a robust framework for
climate risk management. The exercise will test transition and physical risks separately
to focus on exposures directly affected by changes in climate patterns and transition
pathways, based on the scenarios by the NGFS. The physical risk scenario will focus
on the projected climate situation of Hong Kong in the 21st century, such as increases
in temperature, sea levels and more intense cyclones. The transition risk scenarios will
capture disorderly and orderly transition pathways. The financial impact on the banking
sector will be evaluated both in the short-term and long-term. HKMA will allow partic-
ipants flexibility on certain aspects of the test, including reporting granularities and
exercise coverage (HKMA, 2020).
MAS—Singapore
In October 2020, MAS announced that they will begin to incorporate climate-related
scenarios into their annual stress tests for financial institutions over the next two years
(by the end of 2022). MAS’ upcoming stress test aims to assess the impact of physical
and transition risks associated with climate change and references the climate scenar-
ios developed by the NFGS. It is envisioned that the climate stress test will strengthen
the resilience of Singapore’s financial sector to climate-related risks and develop Singa-
pore as a green finance hub. The climate-related scenarios will include both physical
and transition risks, where both qualitative and quantitative information will be incor-
porated. The scenarios will also include a base and a stress scenario, including a short-
term and long-term environmental baseline (MAS, 2020).
APRA—Australia
In September 2021, the APRA published an information paper on its Climate Vulnera-
bility Assessment (CVA), outlining the purpose, design and scope of the exercise. The
exercise has been adapted from existing stress tests with Australia’s five largest banks
taking part. The objectives of the CVA are to measure banks’, the financial system’s and
the economy’s exposure to potential climate risks, understand how a bank can adjust
its business models and undertake management actions in light of the potential risks
faced under different scenarios, and help improve firms’ capabilities in climate risk
management. The CVA is based on two NGFS Phase II scenarios: a delayed but rapid
transition to reduce emissions with high transition risks, and a scenario with limited
further global action with high physical risks associated. The exercise covers a time
period from 2020-2050 using a static and a proportional balance sheet approach. The
aggregated results from the CVA analyses are expected to be published by the APRA in
2022 (APRA, 2021).
RBNZ—New Zealand
New Zealand is taking action to become one of the first countries to mandate
climate-related disclosure for financial entities, in line with the TCFD framework. The
Financial Sector (Climate-Related Disclosures and Other Matters) Amendment Bill
has been introduced to Parliament, making climate-related disclosure mandatory for
around 200 entities. RBNZ also seeks to incorporate climate risks into its stress testing
framework. For its current stress test, it incorporates a disaster shock scenario which
assesses the impact of a natural disaster such as a large-scale earthquake on fiscal
costs and expenditure (RBNZ, 2020).
Below we provide a comparison of eight major climate stress test exercises by supervi-
sory authorities (Table 4).
Table 4: Comparison of key regulatory climate stress tests.
ECB
Danmarks ECB
Economy-
Authority PRA ACPR DNB APRA National- HKMA Supervisory
wide
bank SSM Test
Test
Scope Largest banks Banks and Banks, Banks Banks and Banks Banks Banks
and insurers insurers insurers, companies
pension funds
Inclusion Mandatory Voluntary Voluntary Mandatory Mandatory Not applicable Voluntary Mandatory
Balance- End-2020 End-2019 End-2017 Not disclosed End-2020 Not disclosed Not available End-2021
sheet
reference
Period 30 years (to 30 years (to 5 years (to 30 years (to 30 years (to 9 years 5-30 years Up to 30 years
2050) for 2050) 2023) 2050) 2050) (to 2050),
transition risk depending on
and 60 years the scenario
for physical
risk
Balance- Static Static until Static Static and Static (allows Not disclosed Static Static and
sheet end-2025; proportional for feedback dynamic,
assumption dynamic from loop) depending on
2026 the scenario
Focus of Lending book Business loans, Bond and Mortgages Corporates Corporate Sectoral Mortgages
exposures and large equity, corporate equity and from credit loans and exposures and corporate
corporate credit spreads, holdings and businesses and market mortgages exposure
counterparties sovereign corporate portfolios
credit spreads, loans
commodities
and interest rate
ECB
Danmarks ECB
Economy-
Authority PRA ACPR DNB APRA National- HKMA Supervisory
wide
bank SSM Test
Test
Counterparty- Yes Yes No Yes Yes No Yes Yes
level analysis
Type of risks Physical, Physical and Transition Physical and Physical and Transition Physical and Physical and
transition and transition transition transition transition transition
litigation
Number of Three Four Four Two Three One Not disclosed Five
scenarios
NGFS Built on NGFS NGFS scenarios Not applicable Built on NGFS Built on NGFS Not applicable Based Based
scenarios scenarios used as a scenarios scenarios on NGFS on NGFS
starting point Scenarios scenarios
Calculation Uses internal Uses internal Top-down Bottom-up, Top-down Top-down Uses internal Uses internal
models models stress model top-down approach approach models models
(bottom-up (bottom-up approach approach (bottom-up (bottom-up
approach) approach) approach) approach)
Results due May 2022 Published in Published in Early 2022 Assessment Published in Assessment Will be
May 2021 2018 to be November to be conducted
undertaken in 2020 undertaken in from March to
2022 2021 July 2022
Public Yes, in Yes, in Yes, in Yes, in Yes, in Yes, in Not stated Not stated
disclosures aggregate; not aggregate; not aggregate; not aggregate; aggregate; not aggregate; not
at firm-level at firm-level at firm-level not at firm- at firm-level at firm-level
level
ECB
Danmarks ECB Super-
Econo-
Authority PRA ACPR DNB APRA National- HKMA visory SSM
my-wide
bank Test
Test
Scope Largest banks Banks and Banks, Banks Banks and Banks Banks Banks
and insurers insurers insurers, companies
pension funds
Inclusion Mandatory Voluntary Voluntary Mandatory Mandatory Not applicable Voluntary Mandatory
Balance- End-2020 End-2019 End-2017 Not disclosed End-2020 Not disclosed Not available End-2021
sheet
reference
Period 30 years (to 30 years (to 5 years (to 30 years (to 30 years (to 9 years 5-30 years Up to 30 years
2050) for 2050) 2023) 2050) 2050) (to 2050),
transition risk depending on
and 60 years the scenario
for physical
risk
Balance- Static Static until Static Static and Static (allows Not disclosed Static Static and
sheet end-2025; proportional for feedback dynamic,
assumption dynamic from loop) depending on
2026 the scenario
Focus of Lending book Business Bond and Mortgages Corporates Corporate Sectoral Mortgages
exposures and large loans, equity, equity holdings and from credit loans and exposures and corporate
corporate corporate and corporate businesses and market mortgages exposure
counterparties credit spreads, loans portfolios
sovereign
credit spreads,
commodities
and interest
rate
ECB
Danmarks ECB Super-
Econo-
Authority PRA ACPR DNB APRA National- HKMA visory SSM
my-wide
bank Test
Test
Counterparty- Yes Yes No Yes Yes No Yes Yes
level analysis
Type of risks Physical, Physical and Transition Physical and Physical and Transition Physical and Physical and
transition and transition transition transition transition transition
litigation
Number of Three Four Three Two Three One Not Five
scenarios disclosed
NGFS Built on NGFS NGFS Not applicable Built on NGFS Built on NGFS Not applicable Based Based
scenarios scenarios scenarios used scenarios scenarios on NGFS on NGFS
as a starting Scenarios scenarios
point
Calculation Uses internal Uses internal Top-down Bottom-up, Top-down Top-down Uses internal Uses internal
models models stress model top-down approach approach models models
(bottom-up (bottom-up approach approach (bottom-up (bottom-up
approach) approach) approach) approach)
Results due May 2022 Published in Published in Early 2022 Assessment to Published in Assessment Will be
May 2021 2018 be undertaken November to be conducted
in 2022 2020 undertaken from March
in 2021 2022 to July
2022
Public Yes, in Yes, in Yes, in Yes, in Yes, in Yes, in Not stated Not stated
disclosures aggregate; not aggregate; not aggregate; not aggregate; not aggregate; not aggregate; not
at firm-level at firm-level at firm-level at firm-level at firm-level at firm-level
3.3.3 Importance of internal climate stress tests
As financial institutions mature in their approach to climate stress testing and the
requirements around teams and skills, data and infrastructure, models and scenarios
are increasingly fit-for-purpose, they may find that their risk management and strategy
needs require climate stress tests that differ from regulatory stress testing exercises.
Regulatory stress tests often reflect the concerns and needs of the regulators to deliver
on their supervisory and financial stability objectives, which often differ from the objec-
tives of a financial institution. Below we have listed three main reasons why institutions
may also choose to run an internal climate stress test.
1. Desire to use scenarios and assumptions different from those in the regula-
tory exercise. Institutions may wish to use scenarios and/or assumptions that
are more applicable to their business, for example in relation to plausible transi-
tion pathways, certain policies coming into force, carbon pricing at different time
points, the speed of electrification in a particular industry, the development (or
lack thereof) of carbon capture and storage (CCS). Typically, these would be the
scenarios used for TCFD reporting but institutions may choose to have additional
scenarios for their internal use in planning and risk management.
2. Need for a different or more detailed focus. A regulatory scenario may not
account in enough detail for a particular country or geography, a sector or a
company that the institution believes might carry a specific risk for the institution.
These scenarios might only be used by the institution for internal planning and risk
management purposes.
3. Desire to consider the impacts of specific and sudden events on their portfo-
lios. For example, extreme weather events or a significant shift in public policy
in a particular sector, may have significant and specific impacts on trading book
positions. Market risk scenarios would be a common example of this kind of
impact analysis.
In response, financial institutions can leverage the work done on regulatory climate
stress tests to develop and run a variety of scenarios, at different frequencies, for
different audiences, from comprehensive to very narrow analysis, and all shades in
between (Bank of England, 2021).
4. Team organisation
and skills
Key messages
◾ Institution-wide team engagement is required for climate stress testing, with
firms assigning responsibilities to a large number of teams from across the
organisation for this exercise.
◾ Firms should devote appropriate resources (financial and human) to the execu-
tion of climate stress testing, with a specific focus on the activities of data
collection and analysis, model development and strategic planning.
◾ A large body of knowledge and skills will need to be developed in-house to reach
a satisfactory level of proficiency in running climate stress tests.
◾ Climate-related training and knowledge development programmes need
to be geared towards a diverse set of teams across the firm, rather than just
client-facing employees.
◾ Institutions will need to make substantial changes to their organisation includ-
ing the development of a climate risk team with adequate resources, execu-
tive sponsorship and authority to oversee, coordinate and manage the climate
stress test.
◾ Adequate resourcing should be implemented to integrate climate stress testing
into the firm’s organisational structure and processes with robust governance
and oversight.
Team organisation and skills
4.1 Team organisation
4.1.1 Institution-wide team engagement for climate

stress testing
In order to effectively execute climate stress tests, financial institutions need to
engage a large number of teams from across the organisation. Through UNEP FI’s
TCFD programme, participants suggested the teams that they believe should play
a role in stress testing. Their answers included risk modelling, traditional stress test-
ing, economics, research and relevant business lines. This section will provide greater
detail on the roles and responsibilities required to effectively execute a stress test. In
addition, it will explore the skills needed for both climate stress testing and improved
climate risk management.
This section will also provide specific recommendations for different teams, along with
guidance on how to develop the necessary skills and knowledge. It will also discuss
institutional mobilisation and governance for conducting a climate stress test.
4.1.2 Teams and their respective roles

Figure 4 below illustrates the general functions and the teams that should be involved
in climate stress testing.
Climate risk, Stress testing, Business line, Credit
Climate risk, Stress testing, Research, Business
Risk modelling, Model Validation, Credit risk, risk, Market risk, Operational risk, Country risk,
Climate risk, Research, line, Risk modelling, Model Validation, Credit
Market risk, Operational risk, Country risk and Credit risk and market risk portfolio, Asset
Business line teams risk, Market risk, Operational risk and Country
Asset and Liquidity management teams and Liquidity management and Strategy and
risk teams
Planning teams
Scenario design Financial risk assessment Outcome
Counterparty credit risk
Impact on default
rate
Economic or finan- Macroeconomic
cial shocks impacts
Impact on lenders’ Total impacts on
Climate
earnings banks
Climate related Sectoral & infrasec-
shocks toral impacts
Impact on asset
prices
Liquidity risk
Macro feedbacks
Climate feedbacks
Internal audit and Change Standard modelling structure

Function and IT teams Climate change augmented structure
Figure 4: Functions and teams (adapted from ACPR, 2020).
Table 5 below builds on the above by providing a more detailed overview of the teams that should be involved in stress testing and
their respective roles. As firms differ in their approach to their naming of teams/sub-teams, in order to aid understanding and applica-
tion of these recommendations, we have included a ‘team definition’.
Table 5: Suggested involvement of teams and their responsibilities in climate stress testing.
Team Team Definition Responsibilities

1. Climate risk team ◾ The team whose primary focus is climate risks and their 1. Oversee, coordinate and manage the climate stress testing
(also referred to as impacts. exercise.
Sustainability or ◾ Some financial institutions may not have a dedicated climate 2. Educate other teams involved in climate stress testing.
Environment, Social team and instead have a Sustainability or ESG team that is 3. Act as the central data team to:
& Governance (ESG) responsible for issues related to climate risk.
◽ Collect, clean and validate required data in coordination
team)
with the appropriate teams.
◽ Gather data from open-source databases.
◽ Collect data from third-parties and counterparties.
4. Provide scenario narratives to the scenario design and stress
testing teams.
5. Collate the results and analyse them in collaboration with the
stress testing and risk (credit, market, operational, liquidity
and country) teams.
6. Build a reporting structure for the results.
7. Present the results and analysis at relevant forums.
2. Stress testing team ◾ The team that currently conducts traditional stress testing, 1. Provide the framework of a traditional stress test to incorpo-
including developing and designing scenarios for traditional rate climate stress tests into the framework.
stress tests and strategic planning. 2. Design scenarios based on narratives provided by the
◾ Instead of a single team, the responsibility may be split climate risk team and other teams.
among various risk teams. 3. Undertake scenario expansion.
4. Run the scenarios in collaboration with the risk modelling,
climate risk and other risk teams.
5. Analyse the results of the exercise with the risk teams.
3. Research teams ◾ The teams that conduct economic and climate research. 1. Support scenario design and scenario expansion.
◾ The teams that support the institution’s business and strategy. 2. Remain up-to-date with the latest developments in climate
stress testing methodologies.
4. Business line team The customer-facing division of the financial institution. 1. Assist the climate team in collecting data from clients.
(also referred to as 2. Provide customer insights for counterparty analysis and
front-line) scenario narratives for which their clients will be incorpo-
rated.
3. Provide feedback on the applicability of developed scenarios
for the institution’s clients.
4. Use the results of the stress test to gain customer insights
and inform future relationships.
5. Risk modelling The team responsible for developing the institution’s risk models, 1. Develop climate-specific counterparty models.
(Quants) team including credit risk, market risk and operational risk. 2. Perform sectoral intra- and extrapolation.
3. Play the key role in running and implementing the models
and performing quantitative analysis of the climate risks.
4. Perform tweaks on internal models to improve suitability for
the exercise.
6. Model validation The team responsible for validating models. 1. Validate internal models to be used for climate stress testing.
team 2. Provide expertise and observations to the Quants team to
make the relevant changes to the models.
7. Credit risk team The team responsible for credit risk and internal ratings and 1. Work with the Quants team to incorporate climate risk driv-
methodologies. ers into internal credit models.
2. Collaborate with the stress testing and Quant teams to run
the scenarios.
3. Run the internal ratings tool in coordination with the climate
team.
4. Collate and interpret the results of the exercise from a credit
risk perspective.
8. Market risk team The team responsible for market risk. 1. Work with the stress testing team to include all relevant
markets to be covered in scenarios and feed into the
scenario expansion process.
2. Collate and interpret the results of the exercise to analyse
the market risk results of stress scenarios.
9. Operational risk team ◾ The team responsible for operational risk. 1. Collaborate with the stress testing and climate risk teams to
◾ If assessing litigation risk as part of operational risk, this can design an approach for scenarios to cover the relevant and
also include the legal team. specific operational risks.
2. Collate and interpret the results of the exercise to determine
the operational risk impact of stress scenarios based on the
outputs.
10. Country risk team ◾ The team responsible for country risk. 1. Design country risk methodology to be covered by scenarios.
◾ Generally situated within the credit risk department but 2. Determine country risk impact of stress scenarios based on
collaborating with market risk to gain a comprehensive view the results of the exercise.
of country risks.
11. Credit risk and The teams responsible for aggregate portfolio management of 1. Determine the impact of the results of the exercise on the
market risk portfolio risks. portfolio as a whole.
teams
12. Asset and Liquidity ◾ The team responsible for managing the institution’s assets 1. Determine the impact of the results of the exercise on HQLA,
Management (ALM)/ and liabilities, its liquidity and funding risk (among other the institution’s liquidity, contingency funding plans and fund-
Liquidity and funding duties). ing strategy.
risk team (as part of ◾ Manages high quality liquid assets (HQLA) portfolio at the
Treasury) institution-level.
13. Internal audit team The team that provides assurance on the processes and 1. Audit internal processes for climate stress testing by provid-
controls associated with stress testing. ing an independent assessment of the methodology and
effectiveness of the climate stress testing programme (OSFI,
2009).
14. Change function and The teams responsible for implementing changes in the IT infra- 1. Incorporate the necessary changes in the IT infrastructure to
IT teams structure based on user feedback. allow for processing of climate data for the exercise.
15. Strategy and planning The team responsible for the institution’s strategy and planning. 1. Use the results of the climate stress test to develop or adapt
team the institution’s strategy.
Notes:
The Capital Management Function (usually part of Treasury) also determines Other groups that could have a role may include Economics (perhaps part of
capital impacts and Enterprise Risk Management integrates this into the the Research team) to help with scenario design and translate climate path-
ICAAP including capital and financial plans and risk appetite. This integration ways into macroeconomic and financial variable impact pathways.
would be a key objective in embedding climate risks into an organisation’s risk
framework.
4.1.3 Team resourcing
Discussions on climate stress testing with the UNEP FI TCFD programme’s participat-
ing institutions show that most of the teams involved in climate stress testing only
work part-time on these exercises and have other primary responsibilities.
94% of UNEP FI’s TCFD programme participants have stated that they have at least
one member of their organisation working part-time on climate stress testing.
For example, a credit risk modelling team at an institution works to develop models for
traditional stress tests and internal assessments, and is now, in addition, tasked with
developing climate risk models. The added responsibilities of the execution of climate
stress testing demands the devotion of adequate resources.
Recommendation: Firms should devote the appropriate resources (financial

and human) to the execution of climate stress testing, with specific focus on
the activities of data collection and analysis, model development and strategic
planning.
4.2 Knowledge and skills

4.2.1 In-house capabilities
It is vital for firms to improve their in-house climate risk knowledge and skills,
including those specific to climate stress testing. In the results of its 2020
climate pilot exercise, the Autorité de contrôle prudentiel et de résolution
(ACPR) highlighted the importance of its exercise in improving industry under-
standing of climate risks and their impacts on business models (ACPR, 2021).
It is apparent from the discussion above and regulatory exercises that a range
of climate-related skills are required across a firm. Results from a survey of
UNEP FI’s TCFD programme participants further emphasised the importance
of climate-related skills with participants highlighting the importance of devel-
oping skills, knowledge and expertise on climate scenarios and understanding
the macroeconomic impacts of climate risks. Participants also identified experi-
ence in executing traditional stress tests as a key skill.
Tables 6 and 7 provide an overview of the key climate-specific skills and knowledge
required by teams and functions to ensure the effective design and execution of a
climate stress test. The tables indicate the varying levels of proficiency needed by
the teams for specific skills and knowledge. The table does not include the skills and
knowledge typically required for traditional stress tests, which can be leveraged for a
Knowledge level required
Table 6: Recommended key climate-specific skills required for teams. Basic Intermediate Advanced
Country Risk Team

Climate Risk Team
Liquidity/Funding)
Market Risk Team
Board and Execu-

tive Management
Credit Risk Team
Model Validation
Operational Risk
Business (Client
Quant/modeling
Research Team
Treasury (ALM/
Internal Audit *
Planning Team
Stress Testing
Portfolio Risk
Strategy and
Economic
Teams(s)
Team(s)
facing)
Team *
Team
Team
Key Knowl-
Category edge Short Description
Governance Ability to collaborate with the risk framework owners to integrate climate risk stress testing into
Risk Frame- and Policies existing risk governance and policies.
work Processes and Coordinate the embedding of efficient and effective climated stress tests processes, behaviours and
Controls controls.
Client Engage- Familiarity with various client engagement processes including onboarding, annual review, deals due
ment diligence to assist with the influencing of clients' strategies and sourcing of climate data.
Familiarity with selecting the appropriate metrics for climate risk data to use in a climate stress
Metrics test. For example, for emissions data, the individual should be able to determine which metric, such
selection as total carbon emissions, relative carbon emissions or carbon intesity, is the most suitable for a
Data, Metrics climate stress test.
and Drivers
Experience with collecting physical and transition risk data based on the metrics selected. Individual
Data Collection should be able to collect data from a range of sources, including clients, open-source data sets and
external providers.
Experience with reviewing climate data collected and assessing its coverage, quality, comparability
Data Validation
and production methodology.
Scenario Ability to construct narratives for climate scenarios and to expand reference scenarios to derive the
design and key global, sectoral, country-specific economic indicators which are used in the traditional stress
Scenarios expansion scenario expansion tools.
and Modeling
Experience running climate models, such as IAMs, with an awareness for their assumptions. Ability
Use of Models
to adapt variables for the scenario analysis.
Geospatial Ability to conduct geospatial analysis for physical assets, for example familiarity with the use of GIS.
Analysis Experience in downscaling methodologies.
Analysis
Sectoral Proficient at sectoral intra- and extrapolation for sectoral analysis in a climate stress test, use of
Analysis proxies and estimates where data coverage or granularity is insufficient
Coordination of
Ability to manage and coordinate between various teams, multiple systems and different environ-
multiple teams
ments, without perturbing other functions at the institution
Execution of and systems
the Exercise Producing and
Experience producing the correct desired outputs for the exercise and having the ability to under-
understanding
stand and utilise the outputs.
desired outputs
Reporting and
Ability to define, advise on or challenge the climate risk reports and management information in
Management
support of the governance and management of climate risk, focusing on material areas of risk
Information
Application
of Climate Risk Appetite Ablity to use the outputs of the cilmate stress test to advise on or challenge the metrics and thresh-
Stress Test Setting olds for a fit-for-purpose climate risk appetite.
Results
Sectoral
Ability to devise or advise on meaningful sectoral strategies and policies in light of the stress test
Strategies and
results to support the firm's clients and catalyse the firm's overall climate agenda
Policies
w* Understanding the process and controls around the topics highlighted
Knowledge level required
Table 7: Recommended key climate-specific knowledge areas for teams. Basic Intermediate Advanced
Country Risk Team

Climate Risk Team
Liquidity/Funding)
Market Risk Team
Board and Execu-

tive Management
Credit Risk Team
Model Validation
Operational Risk
Business (Client
Quant/modeling
Research Team
Treasury (ALM/
Planning Team
Internal Audit*
Stress Testing
Portfolio Risk
Strategy and
Economic
Teams(s)
Team(s)
facing)
Team*
Team
Team
Category Key Knowledge Short Description
Climate Understanding of climate Knowledge of climate science, including physical and transition risks, their causes and
Impacts impacts impacts, mitigation & resilience aspects, decarbonisation challenges
Understanding of climate What climate stress tests are and aren’t, how scenarios are constructed, what ques-
stress tests and scenarios tions they can help answer, and what their underlying assumptions are
Details of regulatory and Knowledge of the key features and characteristics of external scenarios, granularity,
international scenarios horizon, main assumptions, severity and coverage
Climate
Scenarios
and Modeling
Climate scenario design and Knowledge of approaches and methodologies for climate scenario design and
expansion expansion
Knowledge and understanding of various models, such as IAMs, including their

Understanding climate
assumptions, how they compare to each other and their integration into a traditional
modeling
stress test framework
What data sources are available, the methodologies used and their assumptions/
Data sources and quality
limitations. Information about coverage, accuracy, granularity
Climate Data
For
Knowledge of the relevant
Knowledge of the relevant climate data and indicators used in the sector/industry their
climate data for each sector/
concerned client/
industry
sector
Understanding of the finan-

Climate risks How climate risks impact economies and the future financial position of firms, the
cial drivers and transmission
and drivers implications for the firms and the institution
channels
For
Sector-specific understand- Understanding of physical and transition risks faced by each specific sector, how they
their
ing of the nature of climate manifest over different time horizons and their significance for each client and their
Sector client/
risks faced by clients sector. Decarbonisation pathways and challenges, mitigation and resilience aspects
& Client sector
Climate
Specifics Sector-specific knowledge
Across the books of the institution, what are the significant climate exposures and
of the key climate risks expo-
associated metrics per sector and geography
sures of the institution
Knowledge on the relevant Awareness of the appropriate regulatory requirements of climate stress testing and
Regulatory
regulatory climate stress test in-depth understanding of the regulator's climate stress test framework, including
Expectations
requirements objectives, data requirements, scenarios, models and their assumptions
* Understanding the process and controls around the topics highlighted
Developing institutional knowledge and skills
As highlighted in the previous section, there is a large body of knowledge and skills that
need to be developed in-house to reach a satisfactory level of proficiency in conducting
climate stress tests, regulatory-related or otherwise.
Set of recommendations to enhance the development of knowledge

and skills for conducting climate stress tests across firms include:
1. Establish staff roles and responsibilities related to climate stress testing.
Clearly include them in job descriptions and objectives of the employees
involved.
2. Develop policies and procedures inclusive of roles and responsibilities:
Ensure that policies are in place to provide sufficient institutional support for
those executing the stress test.
3. Appoint climate risk specialists/ambassadors in each team or function: Drive
effective coordination and collaboration across teams through appointed
climate risk specialists/ambassadors in charge of interacting with the climate
risk team, coordinating the climate stress tests, sharing and disseminating
climate risk intelligence and insights. They may be the same individuals as
the senior experts suggested in greater detail in section 4.3.1.
4. Provide adequate training by experts: Build capacities of specific teams on
their roles and responsibilities for the climate stress testing programme.
Climate-related training and knowledge development programmes are often geared

towards client-facing employees. However, a diverse set of teams across the organisa-
tion need to develop basic to advanced knowledge of the multiple facets of climate risk
and climate stress testing.
Recommendations for firms to consider in their climate knowledge
development programme include the following:
1. Establish a firm-wide programme of climate knowledge development in
alignment with the required level of skills in each area (see section 4.2.1).
2. Tailor the training curriculum to the relevant teams to build their expertise
and capabilities for climate risks and climate stress testing. Knowledge rele-
vant to regulatory scenarios should be prioritised.
3. Set up appropriate training incentives and requirements for existing
resources in the relevant functions.
4. Encourage, support and finance the enrollment of staff in relevant profes-
sional certifications (for example the Global Association of Risk Profession-
als (GARP)’s Sustainability and Climate Risk (SCR) certification, Cambridge
Institute for Sustainability Leadership (CISL)’s sustainable leadership courses
etc.).
5. Engage with climate and data scientists, modelers and tool providers by
organising internal seminars.
6. Establish dialogue and collaboration with local academic and research insti-
tutions, for example to support the development of scenarios with improved
regional granularity and increased applicability for financial institutions.
Outputs from the collaboration should be shared internally and externally.
7. Engage in industry forums and collaborate with peers within the confines
of applicable competition laws and participate in roundtables, seminars and
share best practices with peers.
8. Engage with professional national and international bodies to contribute to
the emergence and communication of best practices.
The above initiatives would fall within the remit of a climate research and intelligence
team, which is described in detail in section 4.3.1. These initiatives can involve senior
experts and ambassadors as relevant.
Exhibit 1: Monetary Authority of Singapore’s Action Plan to build
knowledge and capabilities in sustainable finance.
MAS is supporting Singapore Management University (SMU) and Imperial
College London to open the Singapore Green Finance Centre. The research
centre aims to help institutions and professionals build knowledge and skills
around green finance, including climate finance and climate risk. The initiative
will provide professionals with the necessary skills needed for understanding and
applying climate finance. Programmes will also be offered to undergraduates,
post-graduates and those continuing professional education, to build an under-
standing within the future workforce. The initiative is also supported by nine lead-
ing financial institutions, including Bank of China Limited, BNP Paribas, Fullerton
Fund Management, Goldman Sachs, HSBC, Schroders, Standard Chartered Bank,
Sumitomo Mitsui Banking Corporation, and UBS (Laura Singleton, 2020).
4.2.2 Building skills through external support

A number of participants in UNEP FI’s TCFD programme noted that they had
engaged external support to address skills and resourcing gaps in their climate risk
programmes.
Firms described several sources of external support that are available to institutions
looking to enhance their climate risk programmes and conduct a climate stress test.
These include:
External consultants and contractors can provide expertise on climate risk manage-
ment to support institutions on a range of activities associated with climate stress
testing. External consultants can support data management and integration, support
the development and implementation of climate scenarios, assist in regulatory prac-
tices, and support governance and risk reporting. They are often the preferred route to
respond to regulatory demands and initiate the development of a climate risk function.
National and/or industry associations can help financial institutions comply with new
regulations and understand climate risks and scenarios at the national level. They can
be leveraged to influence policymakers and regulators and more generally facilitate
dialogue with regulators to, for example, clarify points of uncertainty or approaches
to supervision, as well as offer forums for sharing best practices and innovative
approaches.
International initiatives can help financial institutions engage in dialogue with their
peers and leading experts in the industry to expand their understanding of climate risks,
scenario analysis and modelling, as well as address topics such as climate impacts,
data availability and accessibility, risk reporting, best practices and obstacles faced.
External support in the form of consulting, contracting and the use of third-party data
and models continue to be a common choice in the face of rising regulatory demands.
These approaches are a good way to kickstart the process of climate stress testing
and building out a climate risk function. However, external support does bring chal-
lenges including costs, data confidentiality, continuity of the available skillset follow-
ing the conclusion of contracts, the coverage provided, and the ability to transition to
in-house capabilities.
Recommendation: Firms should leverage external support to develop in-house

capabilities for climate stress testing at a faster pace, embed the required skills,
and ensure the ability to produce meaningful climate-related risk analysis and
climate-related risk disclosures.
4.3 Institutional organisation and governance
4.3.1 Changes to team organisation for climate stress testing

Embedding climate stress testing within a financial institution will require changes to
the traditional team organisation of the institution. Below we provide guidance for the
key changes that should be implemented by institutions for climate stress testing.
A traditional stress testing team at a financial institution consists of various risk teams
which share the role of executing the exercise. They oversee key components includ-
ing scenario design and expansion, performing risk analysis and analysing the results.
Other teams, such as the economic research, risk modelling (Quants) and model vali-
dation teams, assist in the exercise by providing their expertise as required. When
the exercise covers several risk exercises, for example, the internal capital adequacy
assessment process (ICAAP) in the UK and EU, the global coordination usually rests
with the firm-wide risk team.
However, a climate stress test requires a different set of skills and knowledge
compared to a traditional stress test, and therefore will require institutions to make
substantial changes to their organisation in order to integrate climate risk assessment
into the traditional stress testing framework.
Recommended changes firms should make include the following:
Develop a climate risk team with adequate resources, executive

sponsorship and authority to oversee, coordinate and manage the
climate stress test
In the case of climate stress tests, a specific team should be established within the
risk department or division with the skillset needed to overcome challenges of the exer-
cise (Figure 5). Depending on the organisation of the firm, the team could be stand-
alone, part of a wider ESG and/or sustainability team or integrated with the firm-wide or
enterprise risk management team.
Climate Risk Team’s Recommended Roles
Oversee, coordinate and manage the climate Educate other teams on climate change, climate
stress testing exercise risks and their impacts
Build a reporting structure for the results and Provide scenario narratives to the scenario
support the governance design and stress testing teams
Collate the results and analyse them in

Present the results and analysis at
collaboration with the stress testing and
relevant forums
risk teams
Figure 5: Recommended roles for the climate risk team.
Some firms may not have a climate risk team overseeing the whole exercise and
instead may be structured as a working group. Though this practice may be suitable
for infrequent stress testing work, it will not be conducive to the adequate integration
of climate stress testing into firm-wide processes and systems.
Establish a climate research and intelligence function

The field of climate risk is rapidly evolving, pushed by the urgency of the climate situa-
tion as well as the scope and amplitude of the changes required. This is particularly the
case in financial institutions, and it is paramount that firms keep abreast of:
◾ The latest climate science;
◾ Regulatory requirements, developments, non-binding standards (e.g. TCFD and
Climate Financial Risk Forum) and reports across jurisdictions;
◾ Reports and initiatives from internationals bodies, agencies and Non-Governmental
Organisations (NGOs);
◾ National and industry associations;
◾ Industry progress and peer analysis; and
◾ Emerging methodologies and best practices, as well as available data sources.
A dedicated climate research and intelligence function can help an institution effec-
tively understand and process a diverse array of climate risk insights. Climate research
and intelligence responsibilities could fit well within the climate risk team, the ESG team
or the sustainability team, depending on the organisational structure of the institution.
Manage climate-related data centrally

Data is a cornerstone of climate risk assessments, analysis and stress testing. Climate
stress testing will require institutions to collect new types of climate-related data that
were not previously required for a traditional stress test. It is crucial that this data
is centrally collected, processed and used to ensure accessibility by all the teams
involved in the exercise. Detailed guidance on data collection is provided in section 5
(Data Requirements and Collection).
Support collaboration among a range of senior experts and establish a
cross-functional project team to design and execute the climate stress test
For a climate stress test, it is very difficult for a single team in an organisation to design,
run, analyse and use the outputs of the exercise on their own. Collaboration of senior
experts from different areas, including business line and risk teams such as climate,
credit, market, operational, liquidity and country risk of the bank is vital for the execu-
tion of the test. This point is further reinforced in the TCFD guidance on risk manage-
ment and disclosure, stating “The principle of interconnections means all relevant
functions, departments, and experts are involved in the integration of climate-related
risks into the company’s risk management processes and in the ongoing management
of climate-related risks” (TCFD, 2020). The need for expertise and knowledge from
across the institution makes cross-functional collaboration crucial. The broader char-
acter of its design and execution requires the involvement of resources and people
from various teams.
The types of teams involved can vary depending on the type of climate stress test
being conducted and the maturity of the organisation in running such processes. The
importance of having a cross-disciplinary team conducting climate stress tests was
also emphasised by the ACPR in their results for their 2020 climate stress test exercise
(ACPR, 2021).
Six recommendations for financial institutions to implement changes

in their organisation for climate stress testing
1. Establish a climate risk team within the risk department, with sufficient
resources and the specific skills and experience required.
2. Establish a climate research and intelligence function in charge of collecting
climate-related information, intelligence, reports and keeping up-to-date with
the latest developments in this field.
3. Assign responsibilities associated with climate-related data to a single team
in the organisation, centralising its sourcing, collection, processing (including
analytics), systems and management.
4. Ensure appropriate dialogue when implementing a climate stress test, all
experts’ opinions should be challenged and checked for consistency. This
process should inform the design and the implementation of the climate
stress tests.
5. Leverage the firm’s implementation of firm-wide stress testing and adapt it
to cater for the specificities of climate stress tests.
6. Clearly and widely communicate on the governance, the key roles and
responsibilities of each team involved in the climate stress testing exercise.
4.3.2 Regulatory guidance on firm-wide integration of climate
stress tests
We highlight below some of the key regulatory recommendations and expectations
when considering the design, implementation and embedding of climate stress tests
with respect to teams and governance.
Embedding climate stress testing into existing stress

testing programmes
In the decade since the GFC, many firms have developed a stress testing programme
that is now strongly embedded within their organisation.
◾ They are designed with clear objectives to meet regulatory requirements, incorporat-
ing views from across the institution.
◾ The programme can include a range of methodologies and models with collabora-
tion from a range of experts.
◾ Recommendations by regulators encourage a well-documented programme, includ-
ing the documentation of “the reasoning and judgements underlying the scenarios
chosen and the sensitivity of stress testing results to the range and severity of the
scenarios” (OFSI, 2009).
◾ Established well-defined internal governance is vital for the effective implementation
of the programme (EBA, 2018).
The firm’s board is held accountable for their traditional stress testing programme
and the senior management is responsible for implementing and managing the
programme and its associated governance.
A climate stress testing programme will additionally require the development of rele-
vant policies and procedures, in order to be well-established across the firm. Given how
strongly traditional stress testing programmes are embedded within financial institu-
tions, it is best for procedures and policies for climate stress testing to be developed
around the current framework for traditional stress testing programmes and its regula-
tory guidance where applicable.
Robust governance and oversight arrangements

Regulators have highlighted the importance of active engagement by the firms’ boards
and senior management in traditional stress testing, as well as in climate risk manage-
ment. A firm’s board is expected to consider climate-related risks in developing internal
strategy and the risk management framework, and to exercise oversight of these risks.
European regulators (ECB, 2020; Bank of England, 2019) have emphasised the impor-
tance of the board and senior management’s understanding of financial risks arising
from climate change, along with their participation in the review and identification of
appropriate, well-understood, documented and sufficiently severe stress scenarios.
For example, in its 2019 Supervisory Statement on enhancing banks’ and insurers’
approaches to managing the financial risks from climate change, the British regulator
stated, “The Prudential Regulation Authority (PRA) expects a firm’s board to understand
and assess the financial risks from climate change that affect the firm, and to be able
to address and oversee these risks within the firm’s overall business strategy and risk
appetite”. Similarly, in its guide on climate-related and environmental risks, the Euro-
pean Central Bank (ECB) detailed its expectations from senior management explain-
ing, “The management body is expected to consider climate-related and environmental
risks when developing the institution’s overall business strategy, business objectives
and risk management framework and to exercise effective oversight of climate-related
and environmental risks.”
These requirements have also been reinforced by the TCFD’s recommendations that
such governance arrangements and results be disclosed (TCFD, 2017), as highlighted
in Figure 6 below. Therefore, the institution’s risk committee or enterprise-wide risk
committee should be responsible for governing the climate stress testing process (and
climate risk management in general), providing effective oversight over the production
of outputs and challenging the associated analyses.
Figure 6: TCFD’s recommendations for governance.
Adequate resourcing for all climate stress testing functions

A generic requirement in all jurisdictions (for example EBA, 2018) is that firms must
ensure that the functions involved in managing risks, including climate-related risks,
have the appropriate human, material and financial resources. Both the European Bank-
ing Authority (EBA) and the Prudential Regulation Authority (PRA) (Bank of England,
2019) have highlighted the importance of board and senior management respon-
sibility for ensuring the required resources and skills are provided for the successful
execution of the exercise. In its 2018 guidelines for institutional stress testing, the EBA
stated, “The management body of the institution should ensure that clear responsibili-
ties and sufficient resources (e.g. skilled human resources and information technology
systems) are assigned and allocated for the execution of the programme” (ECB, 2020).
Integration of climate stress testing into the firm’s organisational
structure and processes
Regulatory guidance available on traditional or climate stress testing refrains from
recommending specific organisational structures as financial institutions differ in their
respective set-ups. However, there are generic structures adopted by institutions that
have been described in section 4.1 as the basis for integrating climate stress testing.
To effectively manage climate risks, the board, executives and the relevant sub-com-
mittees need to have clearly defined roles and responsibilities. For instance, the ECB
expects management at institutions to allocate roles and responsibilities for climate-re-
lated risk management to its members and sub-committees, in accordance with the
three lines of defense model. Similarly, the PRA (Bank of England, 2019) expects there
to be well-established roles and responsibilities for the board and its relevant sub-com-
mittees. The regulator also suggests that the board and senior executives should be in
charge of allocating the responsibility of identifying and assessing climate risks to the
appropriate senior management function (SMF) at the institution.
Exhibit 2: Bank of England (BoE) and Prudential Regulation Authority.

In its 2019 Supervisory Statement, the PRA highlighted the importance of the
involvement of the board and senior management in managing climate-re-
lated financial risks. The regulator expects the board to understand and assess
climate risks by addressing and overseeing these risks within the overall busi-
ness strategy and risk appetite. The PRA specified the need for evidence by insti-
tutions to showcase effective supervision and command by the board and its
relevant sub-committees in climate risk management (Bank of England, 2019).
The PRA has provided further definitive guidance on providing evidence for
governance for their climate stress testing exercise. In its guidance for partic-
ipants for the 2021 Biennial Exploratory Scenario, the Bank of England re-em-
phasised the involvement of senior management, stating, “Participants’
internal governance processes around their Climate Biennial Exploratory
Scenario (CBES) submissions should involve effective challenge from senior
management, including by relevant committees and the board of directors.”
Guidance provided by the regulator on documenting governance included,
“Participants should provide details of these governance... in their responses
to the qualitative questionnaire” and “Participants should include a record of
which committees considered and approved their responses to the exercise and
should also provide board papers relating to their submissions. Participants
should summarise the key issues that were challenged by senior management
or relevant committees, and what changes to responses were made following
this challenge.” As a result, participants are encouraged to disclose details on
the support provided by senior management for governance on the exercise
(Bank of England, 2021).
Nine recommendations for integrating climate stress testing
activities into firm-wide risk frameworks
1. Embed climate stress tests into the institution’s current stress testing
programme, policies and procedures which senior management are respon-
sible for implementing and managing, along with the associated governance.
Recommended governing teams involved: Board of the institution, Board
Risk Committee, executive management teams (CEO, Chief Risk Officer),
senior management functions (Head of Sustainability and ESG, Head of
Risk Functions) and the Climate Risk Working Group.
2. Establish a firm-wide committee for climate risk coordination and gover-
nance. Depending on the structure of the institution, the risk committee or
the enterprise risk committee can take on the role of climate risk coordina-
tion and governance.
Recommended governing teams involved: Risk Committee.
3. Provide the board and senior management with the necessary documenta-
tion. The established firm-wide committee should be responsible for provid-
ing documents that address rationales behind decision-making for the
exercise, implications of choices, key assumptions, limitations of the exer-
cise and evaluation of the results (BIS, 2009).
Recommended governing teams involved: Risk Committee and the
Climate Risk Working Group.
4. Establish and document the roles and responsibilities of the board, its rele-
vant sub-committees and the executive management. The terms of refer-
ence of the relevant committees should be updated accordingly.
Risk Committee, executive management teams (CEO, Chief Risk Officer)
and the Risk Committee.
5. Ensure an adequate allocation of resources for climate stress testing. In
many cases, institutions will need to increase the resources devoted to
climate stress testing.
Recommended governing teams involved: Board of the institution.
6. Increase engagement of the board with senior management on climate risks
to ensure the allocation of resources for climate stress testing and engage-
ment by teams across the institution for the exercise. The infrastructure in
place at the institution needs to be robust but flexible in order to incorporate
changes into the current stress testing programme.
Recommended governing teams involved: Board of the institution and the
senior management functions (Head of Sustainability and ESG, Head of
Risk Functions).
7. Assign roles and responsibilities for the teams and senior management
involved in climate stress testing by the governing authority, leveraging the
existing organisation and arrangements wherever possible.
Recommended governing teams involved: Executive management teams
(CEO, Chief Risk Officer), senior management functions (Head of Sustain-
ability and ESG, Head of Risk Functions) and the Climate Risk Working
Group.
8. Share the outputs of the climate stress tests with the business and present
at the relevant risk committees and to the board, to ensure that the outputs
are embedded into the firm’s strategy.
Risk Committee, Risk Committee and the Climate Risk Working Group.
9. Ensure policies are in place and implemented by senior management for the
adequate use of the outputs of climate stress testing for business strategy
and risk management.
Recommended governing teams involved: Senior management functions
(Head of Sustainability and ESG, Head of Risk Functions).
5. Data requirements
and collection
Key messages
◾ Data needed for a climate stress test can be divided into two main catego-
ries—traditional macro-financial (macroeconomic as well as financial data)
and climate-related data (industry-specific, weather-related, etc)—however,
financial institutions face numerous obstacles when attempting to collect
data for climate stress testing.
◾ For effective climate stress testing, firms should develop policies and proce-
dures to support the collection, processing and use of climate data.
◾ Good practices for collecting data involve identifying data needs, understand-
ing the availability of data sources, implementing industry standards, validat-
ing data, identifying data gaps and adapting institutional systems.
◾ Best practices for securing climate data on and from clients can be divided
into three categories: (i) steps firms can take in-house and through collabo-
rations, (ii) steps requiring client engagement, and (iii) steps to build robust
data collection processes in the future.
◾ For collecting emissions data, firms should identify internal sources through
which emissions data for clients can be accessed and develop internal tools
to calculate emissions data.
◾ When collecting data from external data providers, it is recommended that
firms develop an internal policy, identify easily accessible open-source plat-
forms, develop a questionnaire based on the institution’s data needs and
For climate stress testing, financial institutions must collect climate-related data, in
addition to much of the data required for traditional stress testing. Financial institu-
tions need to collect data to help translate climate risk drivers into financial risks (BIS,
2021). Collecting climate-related data is often a relatively new process for many insti-
tutions and brings with it a variety of challenges. These may include limited availability,
lack of quality, limited granularity and coverage, high costs, interlinking with financial
data, comparability and standardisation, as well as data processing. In the face of
these difficulties, this section provides best practice guidance on in-house and external
data collection by firms for physical and transition risk data.
Data requirements and collection
5.1 Data requirements for climate stress testing
Data needed for a climate stress test can be divided into two main categories: tradi-
tional macro-financial data and climate-related data.
Traditional macro-financial data

Traditional financial data is well-known to institutions that have undergone stress test-
ing after the 2008–2009 GFC. This data includes a variety of financial information
that enables the evaluation of portfolio-level (and sometimes counterparty-level) risk
models. Types of financial data that must be collected for climate stress testing are
described below.
Table 8: Types of traditional macro-financial data required for climate stress testing.
Data type Data required Data Sources

Finance-related ◾ Portfolio composition by sector and geography. ◾ Internal systems
data ◾ External and internal credit and valuation criteria.
◾ Local and global economic data.
◾ Balance sheets.
Macroeconomic ◾ GDP, unemployment, population growth, inflation, ◾ Scenarios
data interest rates and exchange rates.
Data metrics for ◾ Probability of Default (PD). ◾ Internal systems

expected losses ◾ Loss Given Default (LGD).
◾ Average exposure at default (EAD).
Borrower/asset ◾ Costs and revenues by scenario. ◾ Internal systems
financial data ◾ Forward-looking metrics, such as OPEX and CapEx. ◾ Scenarios
Climate-related data
Climate-related data represents additional components that go beyond the traditional
financial data but are required to complete the assessment. Types of climate data that
need to be collected for climate stress testing are described below.
Table 9: Types of climate-related data required for climate stress testing.
Data type Data required Data Sources

Climate hazard ◾ Historical data on acute and chronic physical risks. ◾ Scenarios
data ◾ Projections of future acute and chronic physical risks, ◾ Clients
including their severity and frequency. ◾ External provid-
◾ Adaptive capacity data to determine client resilience ers
and sensitivity to climate hazards, including current
adaptation strategies of clients.
◾ Climate hazard data based on geography, sector and
industry, including economic losses from past climate
hazards.
Data describing ◾ Data on transition risk drivers including policy imple- ◾ Scenarios
transition risk mentation, market shifts, technological changes and
drivers reputation.
Emissions data ◾ Energy and carbon mix of counterparties. ◾ Clients

◾ Published or estimated GHG emissions produced by ◾ External provid-
portfolios and assets of clients. ers
◾ GHG emissions data by region, sector or industry. ◾ Internal systems
◾ Energy efficiency data, for example real estate ratings
like the Energy Performance Certificate Rating.
◾ Data on carbon pricing by jurisdiction.
Climate-related ◾ Identification of the physical assets owned by clients. ◾ Clients
client data ◾ Detailed and granular geographical/geolocational data ◾ External provid-
of assets. ers
Alignment and ◾ Transition pathways set by clients in accordance with ◾ Clients
transition data the Paris Climate Change Agreement. ◾ External provid-
◾ Science-based emission reduction targets set by ers
clients.
◾ Climate policies and pledges of countries.
Regulatory exercises demand the use of both climate-related and financial data in order
to appropriately translate climate risks into financial risks. Exhibit 3 below illustrates an
example from Danmarks Nationalbank on the interlinking of climate and financial data
types to determine the climate risks of lending to emission-intensive sectors.
Exhibit 3: Danmarks Nationalbank interlinking of climate and
financial data.
In its 2020 report “A gradual green transition supports financial stability”,
Danmarks Nationalbank highlighted the results of their sensitivity analyses as
a precursor to a fully developed climate stress test. For their analysis, the regu-
lator used microdata to analyse the impact of climate risks on corporate and
mortgage loans given by banks. For corporate lending, Danmarks Nationalbank
linked accounting data of firms and emissions data at the industry-level to credit
register data for bank lending. For mortgage loans, the regulator linked lending
to energy labels data. Figure 7 below illustrates the linking of bank lending data
to emissions data for emission-intensive sectors to understand the impact of
climate risks across the financial sector (Danmarks Nationalbank, 2020).
Figure 7: Bank lending for emission-intensive sectors.
The difficulties faced by financial institutions in gathering climate-related data as
opposed to finance-related data can be attributed partly to the lack of availability and
understanding of climate data types and financial institutions’ dependence on external
data providers to gather climate data. A survey with participating institutions in UNEP
FI’s TCFD programme concluded that banks face the greatest difficulty in collecting
climate-related client data, emissions data and climate hazard data. The survey results
are shown below in Figure 8. Financial institutions instead find it easier to collect data
on macroeconomic factors, overall portfolio data, data metrics for credit loss (PD and
LGD), borrower/asset financial data and sectoral exposure data.
UNEP FI TCFD participant responses:

Hard to collect data
Emissions data Climate-related client data Climate hazard data
50% 27% 23%
Figure 8: Survey results (2021) showing hardest data types to collect according to
participating institutions in UNEP FI’s TCFD programme.
Common data collection challenges

Financial institutions face numerous obstacles when attempting to collect data for
climate stress testing.
60% of financial institutions in UNEP FI’s TCFD programme have stated that
they are somewhat unconfident or not confident at all in their ability to collect
climate hazard data, with only 4% being quite confident.
Data challenges in identifying and assessing climate risks have also been addressed
by the NGFS in its 2021 report ‘Bridging Data Gaps’ and by the Bank for International
Settlements (BIS) in their 2021 report ‘Climate-related financial risks—measurement
methodologies’. Below we discuss the common challenges faced by institutions in
collecting data for climate stress testing.
Limited Availability
◾ Incomplete disclosure of climate data by clients
◾ Lack of consensus on the climate data required for analysis
Lack of Quality
◾ Absence/limited use of data auditing
◾ Limited disclosure of methodologies/processes underlying data sets
Limited Comparability
◾ Unstandardized data collection approaches
Variability in Granularity and Coverage

◾ Variations in granularity observed at sectoral, country, regional, jurisdictional and firm-
level
High Cost
◾ In-house solutions and external data providers are resource-intensive and costly
Difficulties in Technical Processing

◾ Current IT infrastructure has not been designed to process climate-related data
◾ Need to integrated a new category of systems and processes
Dependence on Third-Party Providers

◾ Data safety requirements can often restrict firms from partnering with a third-party
providers
◾ Third party data can vary in their quality, comparability and regional and sectoral
coverage.
Difficulty Interlinking Data
◾ Limited historical data makes it difficult to determine the relationship between
climate and financial data
◾ Correlation from available historical data may not be reliable enough to determine
future relationships between climate and finance
Figure 9: Key challenges in collecting data for climate stress testing.
Availability. Data availability can be a challenge for two reasons. Firstly, clients do not
generally provide comprehensive and complete climate data disclosures. For exam-
ple, data on clients’ emissions, location of their assets and other forward-looking infor-
mation used to understand the impacts of client transition targets and pathways can
often be sparse. Secondly, the financial sector itself has yet to reach a consensus on
the types of physical and transition risk data required for their analyses.
The lack of availability of climate data was something that the ACPR also highlighted
in its 2020 climate pilot exercise with the regulator emphasising, “the need for better
availability of climate-related data at EU level (physical damage data, transition risks).”
The ACPR further noted that clarificatory questions posed by participating institutions
focused on requests for additional data, especially for sectoral and international expo-
sure (ACPR, 2021).
Exhibit 4: Availability—A key data issue identified by the NGFS.

In its 2021 ‘Progress report for bridging data gaps’, the NGFS identified avail-
ability as one of three main categories in which data gaps exist. The NGFS
found that large data gaps are present in forward-looking data including emis-
sions pathways and transition targets set by firms. Carbon (emissions) data as
well as geolocational data of assets for assessing physical and transition risks
are limited in both their availability and granularity (NGFS, 2021).
Quality. Poor data quality can also present issues for firms conducting climate stress
tests, and regulators such as the De Nederlandsche Bank (DNB) have highlighted the
importance of improving data quality in the context of their stress testing exercises,
highlighting, “sufficiently detailed data to study climate-related risks is not available”
(De Nederlandsche Bank, 2018). Poor data quality is attributable to a range of factors,
including the absence/limited use of data auditing, along with the limited disclosure
of methodologies/processes underlying datasets. Banks, for example, collect data
using dispersed sources but this tends to be subject to limited auditing or independent
validation. Lack of auditing prevents data issues from being rectified, reducing its reli-
ability for climate stress testing. Whilst third-party data may sometimes undergo inter-
polation, extrapolation or other gap-filling processes, this is not always transparently
communicated to data users. The net effect of these shortcomings is that the avail-
able data may not meet the standards set by the institution for a traditional stress test
exercise in terms of coverage, granularity, length and gaps (BIS, 2021).
Exhibit 5: Quality—A key data issue identified by the Basel

Committee on Banking Supervision (BCBS).
In the BIS report ‘Climate-related financial risks—measurement methodologies’,
the Basel Committee on Banking Supervision identified quality of data as one
of the challenges faced of climate-related data. BCBS (2021) stated, “The data
needed to map risk exposures and translate climate-related risks into financial
risk estimates may be only partially available and may not adequately meet
traditional data quality standards, such as the length of history, completeness,
and granularity needed to support the risk decision-maker.” Therefore, lack of
quantity and quality of the climate-data collected by counterparties is a key
obstacle for financial institutions for assessing climate risks (BIS, 2021).
Exhibit 6: Reliability—A key data issue identified by the NGFS.
In its 2021 ‘Progress report for bridging data gaps’, the NGFS identified reliabil-
ity as one of three main categories in which data gaps exist. NGFS explained
that studies have shown that the data sources and metrics available can
produce varying outcomes. A lack of auditing of climate-related data and a lack
of transparency of the methodologies, definitions and criteria used, limits under-
standing of the reliability of the data (NGFS, 2021).
Comparability. Data provided by clients and third-party sources is often not stan-
dardised, making comparisons more difficult both within and between financial institu-
tions. The BIS highlighted this issue in its recent report and noted “Progress has been
less tangible in empirically capturing banks’ exposures to physical risks. This may be
at least partly attributable to considerable additional non-standard data requirements
associated with quantifying physical climate impacts” (BIS, 2021).
Exhibit 7: Comparability—A key data issue identified by the NGFS.

In its 2021 ‘Progress report for bridging data gaps’, the NGFS identified compa-
rability as one of three main categories in which data gaps exist. A lack of stan-
dardisation has led to differences in climate-related disclosure frameworks.
Various methods used to collect, process and link available data have led to a
lack of consistency. A lack of standardisation and a lack consistency can make
it difficult to compare data disclosed using different frameworks due to differ-
ences in definitions and technical standards, such as taxonomies and certifica-
tion labels (NGFS, 2021).
Coverage and granularity. BIS also emphasised the importance of data granular-
ity in assessing physical and transition risks. In its report on measurement method-
ologies for climate-related financial risk, it specifically highlighted the importance of
understanding differences in exposure to such risks at different levels, including at
the sectoral, regional and jurisdictional levels (BIS, 2021). Granularity of data by sector
or geography may vary significantly. For example, whilst large amounts of data are
available for the energy and industry sectors, this tends not to be the case for the IT
or the agriculture sectors (NGFS, 2021). Similar differences in granularity can also be
observed within regions and at the country level. The NGFS considered the challenge
of data granularity in its recent report on data gaps. They identified the fact that differ-
ent types of risks may require different levels of granularity. Table 10 below shows
some illustrative examples of granularity across transition and physical risks.
Type of risk Level of granularity Examples of data
Transition risk Low Country or sector level

Medium Firm-level
High Activity level or value chain
Physical risk Low Country level
Medium District level
High Latitude/longitude
Partly extracted from NGFS, Guide for Supervisors, May 2020.
Figure 10: Data granularity of climate-related risks as identified by the NGFS
(NGFS, 2021).
Exhibit 8: Granularity—A key data issue identified by Basel

Committee on Banking Supervision (BCBS).
BCBS identified granularity of data as one of the challenges faced in relation to
climate-related data, explaining that greater granularity, such as for geolocational
data and climate adaptation data is needed to be able to understand the expo-
sure of small and large institutions to climate risks. Different levels of granularity
among jurisdictions and a lack of standardisation in the data to be collected has
made comparability of data for financial institutions difficult (BIS, 2021).
Resourcing. Acquiring data needs resources and can be costly, for example, in-house
solutions tend to be highly resource-intensive and costly to implement. Whilst banks
often tend to depend on external data providers to obtain climate data, this too can be
costly with firms finding that they have not allocated adequate budgets to purchase
this data nor build in-house data collection capabilities.
Technical limitations. Current bank IT infrastructure and data architecture was not
originally set up for the purpose of processing climate-related data. This can result in
technical difficulties in integrating this data into existing systems. As an example, geolo-
cation tools used to analyse the physical risks associated with a particular set of assets,
represent a new category of systems and processes which will need to be understood
and incorporated into bank data systems along with more traditional client data.
Dependence on third-party providers. Many financial institutions need to use third-
party data providers to access necessary datasets for climate stress testing. This
is partly attributed to many financial institutions lacking necessary expertise and
resources needed in-house for collecting and processing climate-related data (BIS,
2021). Data confidentiality and safety requirements of financial institutions can often
restrict them from partnering with a third-party provider to collect data. Data avail-
able from third-party providers can also vary in quality, comparability, and regional and
sectoral coverage.
Exhibit 9: Limited internal expertise—A key data issue identified
by BCBS.
BCBS identified limited internal expertise and resources within an institution
as a challenge for processing climate data to determine economic impact. The
lack of in-house expertise and resources results in financial institutions relying
on external providers for data (BIS, 2021).
Interlinking of data. A lack of historical data has made it often difficult for financial
institutions to determine the relationship between climate data and financial data for
scenario analysis. Furthermore, correlation from historical data that is available may
not be reliable enough to determine future relationships between climate and finance
(BIS, 2021). When developing scenarios internally, firms have found it challenging to
link traditional macroeconomic variables, such as GDP, to climate-related variables,
such as GHG emissions, to determine the economic impact of climate risks. This was
reiterated in the insights provided in BIS report on stress-testing banks for climate
change risk (2021), stating “historical data are less relevant to climate risk assess-
ments, even where they exist. It is therefore more difficult to estimate linkages between
climate events, climate and environmental policies, the economy, the financial system
and individual institutions.” (BIS, 2021) The obstacle of interlinking climate and finan-
cial data has also been discussed by the NGFS to help bridge data gaps (NGFS, 2021).
Exhibit 10: Use of historical data—A key data issue identified by BCBS.
BCBS emphasised the limitations of historical data of climate risks as historical
relationships between climate and the economy may not be a good representa-
tive of future relationships (BIS, 2021).
5.2 Good practices for collecting internal and

external data for climate stress testing
5.2.1 Good practice steps for data collection

For effective climate stress testing, firms should develop policies and procedures to
support the collection, processing and use of climate data. In the remainder of this
section, we set out good practices in physical and transition risk data collection for
climate stress testing. Good practices involve identifying data needs, understanding
the availability of data sources, implementing industry standards, validating data, iden-
tifying data gaps and adapting institutional systems.
1. Identify data needs
Teams involved in climate stress testing should agree on the data they need, set this as
a data collection target and develop a plan accordingly. To begin, firms should under-
stand regulatory expectations that will influence their data needs. Institutions should
also begin by identifying relevant sectors and regions and the associated data needs.
Once data needs and requirements are determined, teams can begin to identify and
analyse available sources and begin their data collection process. Institutions should
also identify where data gaps exist and develop a plan to improve the scope, quantity
and quality of data collected over time to fill these gaps.
2. Explore a wide range of data sources
Firms should use open-source datasets as a starting point for sourcing climate data,
such as physical hazard data and climate transition scenario pathways. Open-source
platforms help format various datasets into user-friendly formats and can consist of
data based on both past events and forward-looking projections from climate models.
For physical risk data, institutions can also use datasets released by climatological and
geological survey agencies which forecast climate hazards, such as flooding and wild-
fires (BIS, 2021).
After exploring open-source datasets as a starting point, firms should use internally
available datasets, commercial data sources and engage with clients to further collect
the required data for the exercise. A firm’s plan, as described in step 1, should be used
as a guide for collecting data from a diverse set of available data sources to help limit
data gaps and improve quality, granularity and coverage.
Firms should also assess internally available data, such as historical financial data, and
determine whether this can add value to the data requirements for the climate stress
test. For example, access to catastrophe data and loss events may prove useful for the
exercise.
3. Select the most suitable sources
After exploring open and commercial datasets and clients as sources for collect-
ing data, institutions should develop a procedure to compare different data sources
to determine the accuracy and the reliability of the data. By understanding how data
collected through these sources differ, institutions should determine which data
source is the best approximation to use for the exercise.
4. Establish a data validation process
A clear data validation process should be established and documented at the institu-
tion level. The process itself will be dependent on the type of data collected. The source
and methodology used to obtain each data point should be clearly identified. Further-
more, firms should be aware of and document any underlying assumptions made
when using third-party data. When available, overlapping data should be cross-checked
for consistency and be used to inform the collection process, to eliminate potential
unsuitable methodologies or assumptions that lead to poor quality of data.
5. Fill in data gaps with sound assumptions
Data gaps should be identified and the teams involved should agree on how to best
address these gaps. Where no other data is available, institutions should use estimates,
sector averages, weighting and classification schemes within a sector. They may also
choose to use an overlay based on subject matter expert judgement. In order to comply
with regulatory guidance on effective risk data aggregation and risk reporting (for exam-
ple BCBS 239), the institution should develop and document a set of internal proce-
dures and standards to support the acquisition, processing and use of climate data.
6. Develop methodologies to implement policies and procedures for data collection
Standardised metrics and methodologies should be applied for collecting and using
data as data consistency is key to an effective climate stress test. Institutions should
therefore monitor emerging and associated standards for implementation. As a start-
ing point, financial institutions should begin to integrate their data needs into their
internal processes, such as client onboarding, annual review processes, client deals
and due diligence. In its 2021 BEI Client Engagement Guide, the Cambridge Institute
for Sustainability Leadership (CISL) highlighted the importance of open dialogue with
clients during these processes to understand the information the client is publicly
disclosing (CISL, 2021). To ensure standardisation, institutions can develop a question-
naire to ask clients about specific climate risks and establish a framework to deter-
mine the best source for external data on a continuous basis. These internal processes
should be used by the firm to implement policies and procedures designed to improve
climate data collection.
7. Update and adapt internal systems for data processing
Current internal systems used by firms should be updated and adapted to accommo-
date new kinds of data that were previously not required for a traditional stress test.
New data types, such as geographical data (for example longitude and latitude, topo-
graphical data and climate data) should be integrated into the system for processing.
Financial institutions should also incorporate new kinds of systems, such as Global
Positioning Systems (GPS) coordinated for asset location and GIS for mapping asset
location and comparing them to physical hazards projections. Systems on which
models currently run should also be modified, such as current credit systems used in
traditional stress tests.
5.2.2 Guidance on data collection for counterparty analysis

Counterparty-level analysis is important for capturing climate risks for the financial
sector. Climate risk assessment by regulators and financial institutions has centered
around measuring exposure of counterparties to climate risk drivers. This was high-
lighted by the Bank of England in its 2021 CBES exercise where the supervisory author-
ity asked UK banks to conduct analysis on at least 100 of their counterparties, where
exposures were higher than £10 million (Bank of England, 2021).
In order to perform counterparty-level analysis, firms need to engage with their clients
to collect a range of data types to assess physical and transition risks, as described in
Table 10.
Table 10: Climate data required by firms to assess counterparty-specific climate risks.
Type of Type of climate data

climate risk
Physical ◾ Adaptive capacity data, including:
(NGFS, 2021) ◽ Client’s adaptation and resilience plans;
◽ Client’s sensitivity to past climate events, including data on how the
client has tackled extreme weather events in the past.
◾ Geographical data on the location of client’s physical assets.
Note: Hazard forecasts and mapping can be provided by third-parties.
Transition ◾ Alignment and transition plans (e.g. client’s targets and pathways for
(NGFS, 2021) reducing emissions).
◾ Data on a client’s preparedness to transition to a low-carbon economy.
◾ Published or estimated emissions data by jurisdiction, by asset or by
category of assets.
◾ Data available on exposure to carbon prices by jurisdiction.
◾ Energy efficiency and mix data by asset or category of assets.
Recommended guidance for collecting both physical risk and transition

risk data from clients
Given the gaps in climate data disclosure highlighted above, we provide guidance
below on securing climate data on and from clients for the purpose of counterparty
analysis for climate stress testing. This is divided into the following three categories:
(i) steps firms can take in-house and through collaborations, (ii) steps requiring client
engagement, and (iii) steps to build robust data collection processes in the future.
Although this guidance is relevant for securing data from all clients, it is especially rele-
vant for small and medium enterprise (SME) clients. Data on these businesses tends
to be more limited than, for example, large public listed companies, and banks gener-
ally tend to have limited resources available for gathering climate-related data.
Steps for data gathering through in-house and external collaborations

1. Attempt to collect as much relevant climate data on the client. To do so, finan-
cial institutions should use their in-house capabilities to develop internal tools for
data collection. For example, developing a carbon calculator to estimate a client’s
emissions data.
2. Increase stakeholder collaboration by collaborating with various regulators,
governments, municipalities and other stakeholders to increase accessibility to
data, particularly for clients where data is otherwise difficult to collect. For exam-
ple, financial institutions can collaborate with local governments to gain access
to flood assessments at the municipality level, in order to assess flood hazards
as a physical risk in their climate stress testing exercise.
3. Develop industry partnerships to provide the tools and support needed to
clients to facilitate data collection and analysis. Institutions can partner with
various firms, including software companies to leverage the use of digital tech-
nology for data collection. Partnerships can provide financial institutions with
access to cloud and Artificial Intelligence (AI) technology which can be used to
better inform their understanding of their client’s data by collecting large amounts
of data consistently at a reasonable cost.
Recognising the importance of AI technology for data collection, a new initiative
known as OS-Climate, was launched by the Linux Foundation, in collaboration
with Amazon, Allianz, Microsoft and S&P Global, channeling AI technology with
open-source analytics and open data for assessing climate risks.
Steps requiring client engagement

1. Facilitate open and effective communication with the client about data require-
ments. Institutions should take steps to increase communication with their
clients to ensure they understand data requirements and that they disclose
the correct data. Effective communication on data collection can be facilitated
through workshops, courses and materials to build clients’ understanding in this
area. Open communication will also help the financial institution to understand
the client’s current progress in disclosing data, their capabilities and the obstacles
they face in data production. This will mean firms are better placed to support
clients going forward.
2. Update current client engagement processes such as underwriting, know your
customer (KYC) and other due diligence processes, and integrate the climate
stress test data requirements. Data collection from clients should be set-up to be
made as painless as possible to ensure the data needed is disclosed by clients.
Steps to build robust data collection processes in the future

1. Develop a questionnaire based on the data needed from the client. The ques-
tionnaire should act as a guide for financial institutions and clients to ensure that
necessary information is collected for a climate stress test. Detailed examples of
the types of questions that should be asked in the questionnaire are included in
Appendices and should vary depending on the data type. A questionnaire will help
financial institutions overcome challenges related to data availability, comparabil-
ity, coverage and quality. The questionnaire can be provided during the onboard-
ing process, annual review, due diligence process or ad-hoc climate related
discussions. Appendix 1 provides examples of key questions which can serve as
a mini questionnaire.
2. Adopt the use of GIS to gather geospatial data related to clients and their assets,
to identify at-risk locations. New kinds of systems and data will require firms to
upgrade their IT infrastructure.
See Appendix 1 for key questions and topics to address when collecting physi-
cal risk data from clients
See Appendix 2 for key questions and topics to address when collecting transi-
tion risk data from clients
Figure 11: Examples of low-carbon transition initiatives that clients can participate in.
Exhibit 11: Counterparty data guidance by the Bank of England.

For participants of the 2021 Climate Biennial Exploratory Scenario, BoE has
provided in-depth guidance with examples of when corporate counterparty
adaptation plans should be included into climate stress testing analysis by insti-
tutions. Below we have summarised the guidance provided by the supervisor.
Examples where corporate counterparty adaptation plans should not be
included in the analysis include where the counterparty:
◾ Does not have an adaptation plan or is not implementing an adaptation plan.
◾ Is implementing an adaptation plan but it is not likely to be completed.
◾ Belongs to a regulated industry with upcoming regulatory changes for transi-
tion, but the adaptation plan is not currently being implemented.
Examples where counterparty adaptation plans might be included in the analy-
sis include where the counterparty:
◾ Has adaption plans, began implementing them before the end of 2020 and
the plans are highly likely to be completed.
◾ Is implementing long-term adaptation plans, which they began implementing
before the end of 2020.
Examples where counterparty adaptation plans might be broken down and only
certain parts are included in the analysis include where the counterparty:
◾ Has implemented adaptation plans that are highly likely to be completed and
include unachievable long-term targets.
◾ Has implemented certain parts of their adaptation plans but other parts
should not be accepted.
(Bank of England, 2021)
Additional guidance on collecting emissions data

63% of institutions participating in UNEP FI’s TCFD programme ability. are either some-
what or not at all confident in their ability to collect emissions data, with only 4% of the
institutions extremely confident in their ability.
Below we have provided additional recommendations for collecting emissions data for
a climate stress test
1. Identify internal sources through which emissions data for clients can be
accessed. For example, firms can calculate emissions data by accessing electric-
ity consumption payments made by the client through them.
2. Develop internal tools to calculate emissions data, such as a carbon calculator.
A carbon calculator is a useful and easy-to-use tool to estimate a client’s GHG
emissions and determine their carbon footprint by using information from their
banking accounts related to their electricity, gas and fuel payments. Simple tools
like a carbon calculator do not require additional information from clients.
Exhibit 12: Example—GHG Emission Calculation Tool by the

Greenhouse Gas Protocol.
The GHG protocol launched a free to use and publicly available GHG Emission
Calculation Tool. The excel-based tool was developed in partnership with the
World Resource Institute (WRI) to help calculate companies’ GHG emissions.
The tool uses default emission factors but also allows users to provide custom
emissions factors as shown in Figure 12 below. Users are also able to adjust
the default global warming potential metric.
Figure 12: Custom emissions factors for companies
The carbon calculator allows the calculation of Scope 1, 2 and 3 emissions. The
tool calculates stationary combustion, mobile combustion and refrigerants as
Scope 1 emissions, purchased electricity as Scope 2 emissions and transporta-
tion as Scope 3 emissions. Below shows the methods used by the tool to calcu-
late emissions.
Figure 13: Example of Scope 1 emissions calculations.
Below is an example of the emissions summary provided for a company by the
excel tool.
Figure 16: Example GHG emissions summary for a company.

(Greenhouse Gas Protocol calculation tool, 2021)
3. Attempt to collect emissions data reported by a client in their environmental

reports and official fillings. The Partnership for Carbon Accounting Financials
(PCAF) (PCAF, 2020) recommends the use of the most recent data disclosed by
companies in their official filings and environmental reports.
4. Develop an additional questionnaire focused on published and estimated GHG
emissions for companies. The questionnaire should cover methodology, sources
and the shortcomings of the data on emissions. In the appendices, we have
provided examples of key questions which can serve as a mini questionnaire.
See Appendix 3 for additional questions to include when collecting emissions
data from companies.
5. When emissions data is not available from clients, especially SME clients, firms
should use sectoral or industry averages and assign them to the client, with the
possible overlay of subject matter expert judgement as to the specificities of the
client. PCAF encourages the use of estimation models to determine GHG emis-
sions emitted by a company when they do not disclose their data. For example,
PCAF suggests the use of emission factors from production-based models and
revenue-based models to fill in data gaps (PCAF, 2020).
6. Collaborate with various groups and stakeholders to collect emissions data for
companies. For example, financial institutions can collaborate with grid regula-
tors to obtain the energy usage of a certain local area.
5.2.3 Types of third-party data available

Financial institutions also need to buy from or partner with external data providers to
access available data, to meet climate stress test data requirements, as shown previ-
ously in Table 9. In its guidance for climate stress testing, the BoE provided the partic-
ipating firms with benchmark physical risk data from external data providers such as
the UK Met Office, NGFS and the Oasis Hub. Below we have highlighted the data types
needed for climate stress testing, which financial institutions can obtain through part-
nerships with external providers (Table 11). Data from external providers can either be
open data sources, as described in section 5.2.4, or commercial data sources. Data
can also be obtained from scenarios and models, which will be discussed in section 6.
Table 11: Climate data accessible through third-party providers
Type of climate Type of climate data

risk
Physical ◾ Past and future projections of frequency and severity of acute and
chronic physical events, including physical risk hazards at the:
◽ Regional level;
◽ Sector-level; and
◽ Industry-level.
◾ Geolocation of assets.
Transition ◾ Country-level data on carbon pricing.

◾ Carbon footprint and GHG emission data at the:
◽ Firm-level;
◽ Country level;
◽ Sector-level; and
◽ Industry-level.
◾ Pledges, targets and commitments announced by firms and countries.
Exhibit 13: Internal data collection by the European Central Bank.
In March 2021, the ECB released the results of its internally conducted climate
stress testing exercise. For the exercise, the regulator constructed a dataset by
combining financial and climate data for four million global firms, as illustrated
in the Figure 17 below. They used external services to collect data, deriving
data from 427 datasets for physical risks and Urgentem datasets for emissions
data. Emissions data was collected for these firms to determine the impact of
climate policies. The emissions data included past and future emissions taking
into account reduction targets set by the firms. ECB also used geolocation data
for matching potential climate hazards to the physical address of the firm to
calculate a physical risk score for each firm. Various data sources were used to
interlink climate data with financial variables for analysis.
The datasets generated by the ECB helped highlight the difference in vulnera-
bility to climate risks among sectors and regions, emphasising the importance
of granularity of the data collected. The firm-level data were combined with the
NGFS scenarios to identify and quantify climate risk exposure of firms globally.
Figure 17: Key features of the ECB climate stress.
(ECB, 2021)
Gathering data from third-parties
Recommendations for gathering physical and transition risk data

from external data providers
1. Develop an internal policy for gathering climate risk data from external data
providers. The recommendations below should be considered when develop-
ing these policies.
2. Identify open-source platforms that can provide access to climate-related
data, including climate hazards data and transition risk data. For example,
the GHG Protocol (World Resources Institute & World Business Council for
Sustainable Development, 2004) recommends the use of GHG registries
which have been established as open databases for firms to report their GHG
emissions. These registries are often established by governments, NGOs, or
industry groups.
3. Identify potential data providers to work with. Institutions should be fully
aware of the suitable data providers available in their local region, with whom
they can collaborate with.
4. Develop a questionnaire based on the institution’s data needs from the data
provider. The questionnaire should be designed to determine coverage and
methodologies used by the providers of interest to the institution. Institutions
should incorporate the questionnaire into their selection process for exter-
nal data providers as a guide to determine which provider or database best
matches the data requirements for the climate stress testing exercise. In the
Appendices, we have provided examples of key questions which can serve as
an initial questionnaire.
5. Actively communicate with data providers and determine which provider to
collaborate with or buy from. Financial institutions should take part in initia-
tives designed to assist institutions in their data collection for climate risk
management from a range of different data sources. Dialogue on data collec-
tion can be initiated through initiatives such as the Future of Sustainable Data
Alliance.
See Appendix 4 for key questions to address when gathering physical risk data
from external providers
See Appendix 5 for key questions to address when gathering transition risk data
from external providers
Third-party public data sources for physical risk data.
Financial institutions should access and analyse all open source/public platforms to
collect data on physical risk hazards free-of-charge. Below is a non-exhaustive list of
publicly available global data sources on physical risk hazards.
Table 12: Open data sources available for physical risk hazards.
Data Source Physical hazards covered Geographical

Coverage
Climate Central Extreme sea levels, storm surge data, high tide Global
events, coastal flooding, sea level changes and
severe winds.
PREPdata Temperature rise, precipitation, coastal risk, water Global coverage with low
risk and other extreme events. granularity for specific
countries
UNEP Global Tropical cyclones, storm surges, drought, earth- Global
Risk Data quakes, fires, floods and landslides.
Platform
KNMI—Climate Temperature, droughts, cyclones and precipitation. Global
Explorer
World Bank Temperature rise, seasonal precipitation, sea level Global
Climate Change rise, extreme weather events, such as floods,
Knowledge droughts and heat waves.
Portal
WRI Aqueduct Water risks, including flood and drought risk. Global
Water Risk
Atlas
CDP Open Data Storms, extreme heat, sea water intrusion, drought, Global coverage of CDP
Portal floods and forest fires. cities
GFDRR Extreme heat, floods, earthquakes, landslides, sea Global

ThinkHazard! level rise, water scarcity, and wildfires.
Oasis Hub Flooding, cyclones, earthquakes, extreme weather Global

and landslides.
Open Data for Flooding and cyclones. Estimated 55 countries
Resilience Index covered
Google dataset Hurricanes, sea level rise and temperature rise. Global
search
Third-party public data sources for transition risk data

Financial institutions should access and analyse all open-source platforms to collect
data on transition risks free-of-charge. Below is a non-exhaustive list of publicly avail-
able data sources on transition risks.
Table 13: Open data sources for emissions data
Data Source Transition Risks Covered Geographical Coverage

CAIT Climate Data GHG emissions, emission path- Global
Explorer (by WRI) ways, pledges and targets.
CDP Open Data GHG emissions. Global

Portal
The Carbon GHG emissions estimates for United States of America, European
Monitoring power plants. Union, Canada, India, and South
for Action (CARMA) Africa, as well as data from the
database International Atomic Energy Agency
Greenhouse Gas Product life cycle and corporate Global

Protocol value chain (Scope 3) GHG inven-
tories.
The Lowdown v2.0 Coal capacity for countries. Global
UNdata Methane, Carbon Dioxide, Hydroflu- 43 countries however data is only

orocarbons, Nitrous oxide, Nitrogen available for 29 years
trifluoride, Perfluorocarbons and
Sulphur hexafluoride.
6. Models and scenarios
Key messages
◾ Adapt reference scenarios geographically and sector-wise and derive the
impact on key drivers to develop a relevant scenario narrative.
◾ Implement a process to integrate the latest scenario developments and
emerging methodology standards into the scenario design for climate stress
testing.
◾ Undertake scenario expansion to extrapolate the additional scenario vari-
ables required for the climate stress test.
◾ The modelling process in a climate stress test consists of modelling climate
variables, measuring the impact of climate risks on macroeconomic vari-
ables, breaking down the macroeconomic impacts per sector or per portfolio
and quantifying the impact on the financial institution.
◾ To improve modelling capabilities, a firm should identify gaps in current
in-house models and increase communication with external modelers and
academics on scenario expansion and improve understanding of different
models. Firms should develop criteria to enhance their coordination with
these groups.
◾ Adapt existing downstream models, for example loss forecasting models, to
incorporate climate-related variables
6.1 Climate scenarios for stress testing
6.1.1 Overview of climate scenarios

In the financial sector, scenario analysis is a commonly used tool by risk managers. A
scenario provides an alternative state that is plausible in the future under a given set
of assumptions and constraints. In its 2017 technical guide on scenario analysis, the
TCFD defined climate scenarios as:
Models and scenarios
“A path of development leading to a particular
outcome. Scenarios are not intended to represent a
full description of the future, but rather to highlight
central elements of a possible future and to draw
attention to the key factors that will drive future
developments”
TCFD, 2020
Climate scenario analysis has been growing in importance for climate risk assessment,
especially for climate stress testing.
22% of UNEP FI’s TCFD programme participants are using climate scenarios for
their climate stress testing exercises.
In a traditional stress test, scenarios are used to understand the potential impact of
a stress event on a financial institution using macroeconomic variables. In a climate
stress test, scenarios include physical and transition risk variables, which are trans-
lated into macroeconomic variables to understand the potential impact of climate
stress events on a firm. The physical and transition risks included in the scenarios are
those that are likely to have a material impact on a financial institution. The Financial
Stability Institute’s report “Stress-testing banks for climate change—a comparison of
practices” further explained that scenarios “need to take into account cumulative and
feedback effects arising from both sets of risks and economic impacts.” A climate
stress test can require a higher number of scenarios due to uncertainty in the physical
and transition risk variables used. A greater number of scenarios produces a greater
range of outcomes to help financial intuitions understand the potential impacts of
climate risks (BIS, 2021). Below we describe some of the common climate scenarios
used for scenario analysis and climate stress testing.
Network for Greening the Financial System reference scenarios

In June 2020, the NGFS, a global group of central banks and supervisors, released its
Phase I of climate “reference” scenarios. The NGFS climate scenarios examine various
futures in relation to climate change—an orderly transition to a low-carbon economy,
a disorderly transition and a ‘hot house’ no action world. Each reference scenario has
a storyline with a set of characteristics, such as temperature targets or deployment
of carbon dioxide removal (CDR) technologies. Figures 18 and 19 show the classifica-
tions and storylines of the NGFS scenarios.
Strength of response
Based on whether climate targets are met
Met Not met
Disorderly Too little, too late

Sudden and unanticipated We do not do enough to
Disorderly
response is disruptive but meet climate goals, the

sufficient enough to meet presense of physical risks
climate goals spurs a disorderly transition
Transition pathway
Transition risks
Orderly Hot house world

We start reducing emis- We continue to increase
sions now in a measured emissions, doing very little,
Orderly
way to meet climate goals if anything, to avert the

physical risks
Physical risks
Figure 18: Classification of the NGFS Reference Scenarios.
Scenario Classification Storyline

Current policies Hot house world Current climate policies are followed for all regions, similar to
IEA’s current policies scenario, implies rise of approx 4°C by
2100
Nationally deter- Hot house world Nations follow unconditional nationally determined contribu-
mined contributions based on the Paris Agreement, implies rise of approx 3°C
tions (NDCs) by 2100
Immediate 2°C Orderly transition Collective action is taken now to reduce emissions towards a
2°C target, very similar in structure to the old 2°C scenarios
Immediate Orderly transition Collective action is taken now to reduce emissions towards a
1.5°C 1.5°C target, very similar in structure to the old 1.5 scenarios°C
Immediate 2°C Orderly transition Agressive collective action begins now on 2°C pathway, limited
(low CDR) use of negative emissions
Delayed 2°C Disorderly tran- Agressive collective action only begins after 2030 to align with
sition a 2°C
Delayed 2°C Disorderly tran- Agressive collective action only begins after 2030 to align with
(low CDR) sition a 2°C, limted use of negative emissions
Immediate Disorderly tran- Agressive collective action begins now on 1.5°C, limited use of
1.5°C (low CDR) sition negative emissions
Figure 19: Storylines of the Phase I NGFS Reference Scenarios.
In June 2021, the NGFS released Phase II of its climate reference scenarios. These
scenarios have been improved and updated to capture the latest developments in
climate data and analysis. The scenario narratives now include regional policies to
reach net zero by 2050 to limit warming to 1.5°C. Phase II has six scenarios instead
of the eight scenarios from the previous phase and still sits within the NGFS scenario
framework of orderly, disorderly and hot house world pathways. Figure 20 highlights
the key changes in scenarios from Phase I to Phase II by the NGFS.
Phase I name Phase II name Changes

◾ Orderly (1.5°C with ◾ Net Zero 2050 This scenario still reaches 1.5°C and net zero
CDR) by 2050. Individual regional pathways are
updated.
◾ Orderly (rep) (2°C with ◾ Below 2°C This scenario still leads to warming between
CDR) 1.5–2°C. CDR has been limited in the new
◾ Orderly (2°C with scenario.
limited CDR)
◾ Disorderly (1.5°C ◾ Divergent Net-Zero This scenario now reflects the impact of
limited CDR) Policies divergent policies across sectors and regions
represented as carbon price variation.
◾ Disorderly (rep) (2°C ◾ Delayed 2°C Delay to policies still occurs until 2030. CDR
delay with limited has been limited in the new scenario. In
CDR) addition, the scenario includes regional carbon
◾ Disorderly (2°C delay price variation, i.e. regions with net-zero targets
with CDR) are more ambitious than regions without them
after 2030
◾ NDCs ◾ NDCs Emission and temperatures are lower due to
increased climae policy commitments
◾ Current Policies (rep) ◾ Current Policies Overall emissions and temperature increase is
lower due to lower baseline growth assump-
tions and more policies already implemented.
Figure 20: Comparison of Phase I and Phase II NGFS Scenarios.
International Energy Agency (IEA)
The IEA has developed numerous scenario energy projections for the medium and the
long-term using its World Energy Model (WEM) (IEA, 2020). The transition scenarios
developed by the IEA represent various policy outcomes, temperature rise and energy
and economic pathways.
Stated Policies Scenario (STEPS)

STEPS provides a detailed picture of the direction of current policy initiatives and ambi-
tions and their impact on the energy sector through to 2040, exploring climate-related
transition risks. Considering existing policies and commitments impacting the energy
market, the scenario aims to provide a benchmark to assess the potential achieve-
ments and limitations of developments in energy and climate policies. The pathway
includes decreased renewable costs, changes in the global reliance on oil, and the
effects on demand, prices and emissions predictions. (IEA, 2020).
Sustainable Development Scenario (SDS)

SDS maps out a pathway that requires rapid and widespread changes across all
aspects of the energy sector to meet sustainable energy goals through to 2050. The
SDS holds temperature rise to below 1.8°C with a 66% probability and takes into
account the impact that investment may have on supply-side and demand-side
processes, such as to ensure a reliable energy supply and allow firms to take advan-
tage of the vast potential for energy efficiency (IEA, 2020; IEA, 2021).
Net-Zero Emissions by 2050 (NZE2050)

The NZE2050 scenario shows what more is needed to be done beyond the SDS path-
way for the global energy sector to be on a pathway to net-zero emissions by 2050. The
scenario is in line with the current pathways used by the IPCC and includes detailed
modelling of what would be needed over the next ten years to put CO2 emissions on
track for global net-zero emissions by 2050. The scenario places less emphasis on
carbon capture, utilisation and storage (CCUS) and more emphasis on low-carbon,
renewable energy sources that can supplement the scenario’s objectives (IEA, 2020).
Intergovernmental Panel on Climate Change IPCC

In its Fifth Assessment Report published in 2013, the IPCC developed scenarios to
project possible future pathways relative to radiative forcings. Radiative forcings
measure the combined effect of GHGs and other factors which increase the trapping
of additional heat due to climate change. Each scenario, known as a Representative
Concentration Pathway (RCP), shows one of the many possible pathways for a certain
level of radiative forcing (Carbon Brief, 2019).
RCP 2.6
RCP 2.6 scenario pathway was developed by the IMAGE modelling team of the PBL
Netherlands Environmental Assessment Agency and aims at limiting the global
temperature rise to likely below 2°C (IPCC, 2021). It is a stringent mitigation scenario
that requires a major turnaround in climate policies and concerted climate action glob-
ally (Van Vuuren et. al., 2011). The pathway spans from 2010-2100, with CO2 emissions
peaking by 2020 but significantly declining over time (RCP Database, 2021).
RCP 4.5
Developed by the GCAM modelling team at the Pacific Northwest National Laborato-
ry’s Joint Global Change Research Institute (JGCRI) (Van Vuuren et. al., 2011), RCP 4.5
is one of the intermediate scenarios used by the IPCC (IPCC, 2021). It assumes that all
countries undertake emissions mitigation simultaneously and effectively and share a
common emissions price (Thomson et al., 2011). In RCP 4.5, the pathway sees emis-
sions peak mid-century at ~50% higher than 2000 levels (Van Vuuren et. al., 2011).
RCP 6
Developed by the Asia-Pacific Integrated Model (AIM) team at the National Institute for
Environmental Studies (NIES) (RCP Database, 2021), RCP 6 is one of the intermedi-
ate scenarios used by the IPCC. In this scenario emissions double by 2060 and then
dramatically fall due to use of innovations in the energy sector and assumptions of
lower technology costs, but remain well above current levels. CO2 concentration contin-
ues increasing, though at a slower rate in the latter parts of the century (IPCC, 2021).
RCP 8.5
This scenario is the worst-case scenario in which emissions continue to increase rapidly
through the early and mid-parts of the century. Developed using the MESSAGE model
and the IIASA Integrated Assessment Framework by the International Institute for
Applied Systems Analysis (IIASA), in this scenario, emissions continue to increase rapidly
through the early and middle of the century. The pathway represents no climate change
mitigation, therefore no mitigation measures are implemented (Carbon Brief, 2019).
Table 14 highlights how climate scenarios from different modelers can vary in their
features and assumptions. The table compares the NGFS scenarios to other climate
scenarios, such as those by the IEA and the IPCC.
Table 14: Comparison of NGFS scenarios to other climate scenarios.
Feature NGFS Phase II Scenarios Other Scenarios

Timeframe ◾ Up to 2100 for climate impacts and ◾ RCPs (IPCC) go up to the year 2100.
transition pathways. ◾ NZE2050 (IEA) goes up to the year
◾ Up to 2050 for economic impacts 2050.
(NiGEM).
Scenario ◾ Net Zero 2050—limits warming to ◾ NZE2050 (IEA)—limits warming to
Risk Drivers 1.5°C 1.5°C.
and ◾ Below 2°C—limits warming to 1.7°C. ◾ SDS (IEA)—limits the temperature rise
Level of ◾ Divergent Net Zero—limits warming to to below 1.8°C with a 66% probability
Policy 1.5°C. without reliance on global net-negative
Ambition CO2 emissions.
◾ Delayed transition—limits warming to
1.8°C. ◾ RCP 2.6 (IPCC)—temperature rise is
not likely to exceed 2°C.
◾ Nationally Determined Contributions
(NDCs)—limits warming to 2.5°C. ◾ RCP 4.5 (IPCC)—temperature rise
more likely than not to exceed 2°C.
◾ Current Policies—warming to occur up
to 3°C or higher. ◾ RCP 6.0 (IPCC)—temperature rise likely
to exceed 2°C.
◾ RCP 8.5 (IPCC)—temperature rise as
likely as not to exceed 4°C.
Policy ◾ There is a large variation in the policy ◾ STEPS (IEA) scenario considers all
Reaction (carbon price) depending on the existing or announced carbon pricing
scenario. schemes.
◾ A higher emissions price indicates ◾ SDS (IEA) scenario assumes that
more stringent policies. carbon pricing is established in all
◾ For the net zero by 2050 scenario, developed economies and reach the
IAMs suggest a carbon price of price USD140/ tonne in 2040.
around USD160/ tonne by the end of
the decade.
◾ This price indicates overall policy
intensity. In reality, governments are
pursuing a range of fiscal and regula-
tory policies.
Technology ◾ Net Zero 2050 and divergent net-zero ◾ NZE2050 (IEA)—majority of reductions
Change scenarios assume fast technological in CO2 emissions up to 2030 are due to
changes. technologies currently available. Half
◾ Below 2°C scenario assumes a moder- of reductions in CO2 emissions 2050
ate change. come from technologies currently
being developed.
◾ A delayed transition scenario assumes
slow/fast change.
◾ Nationally Determined Contributions
(NDCs) and current policies scenarios
assume a slow change.
Feature NGFS Phase II Scenarios Other Scenarios
Carbon ◾ The NGFS scenarios assume low to ◾ In the IPCC RCP scenarios, CO2
dioxide medium availability of CDR technolo- emissions captured and stored from
removal gies. bioenergy and direct air capture range
◾ Afforestation is heavily dependent on from 3.5-16 Gt CO2 in 2050.
(CDR)
the scenario and is high for MESSAGE ◾ In the IPCC RCP scenarios, land use
under all scenarios. varies from net reforestation to defor-
estation.
Population ◾ Scenarios are based on Shared Socio- ◾ IEA (WEM)—population growth is
Assumption economic Pathway (SSP) 2. based on United Nations Popula-
◾ SSP 2 assumes that society evolves tion Division report (UNPD, 2019).
in line with past trends and the global In WEO-2020, world population is
population will peak around 2070. projected to grow from 7.7 billion in
2019 to 9.2 billion in 2040.
◾ IPCC RCP scenarios do not include
socioeconomic narratives.
Economic ◾ Scenarios are based on SSP 2 . ◾ IEA (WEM)—For the short and
Assumption ◾ The GDP pathway was adjusted to medium-term, economic growth
take the COVID-19 pandemic into assumptions are in line with the latest
account. estimates from the IMF. In the long-
term, growth across various regions is
assumed to convert to an annual rate,
with the effects of COVID-19 taken
into account.
◾ IPCC RCP scenarios do not include
socioeconomic narratives.
Models ◾ Climate models participating in the ◾ IEA scenarios use the World Energy
Used ISIMIP project: Model (WEM).
◽ CLIMADA ◾ IPCC Fifth Assessment Report (AR5)
◽ REMIND-MAgPIE featured climate models of the
Coupled Model Intercomparison Proj-
◽ GCAM
ects (CMIP) 5.
◽ MESSAGEix-GLOBIOM 1.1
◾ IPCC Sixth Assessment Report (AR6)
◽ NiGEM featured models from CMIP 6.
◽ IAMs
6.1.2 Regulatory scenarios and the NGFS

For supervisory climate stress tests, most supervisors have leveraged NGFS reference
scenarios from Phase I and II and have expanded from those for their climate scenario
analysis. Reference scenarios need to be adopted by regulators and firms to increase
the relevance of the scenarios to geographies, markets and financial systems (IFF, 2021).
Survey responses show that 100% of UNEP FI’s TCFD programme participants
feel that under stress conditions, the NGFS scenarios assess a portfolio only
slightly or somewhat well.
Therefore, it is important for scenario users to adapt reference scenarios for their institution.
Below we have provided examples of supervisory exercises that have used or plan to
use NGFS scenarios as a foundation.
◾ Australian Prudential Regulation Authority (APRA): For its Climate Vulnerability
Assessment, APRA has designed its scenarios based on two of the NGFS Phase II
scenarios. The first scenario aligns with the NGFS delayed transition scenario, and
the second scenario aligns with the NGFS hot house world: current policies scenario,
with the regulator stating, “Nevertheless, there will be some differences that arise in
the implementation of the NGFS scenarios, due to the limited sector and regional
resolution of the scenarios as well as the selection of models within the NGFS
scenarios” (APRA, 2021).
◾ Hong Kong Monetary Authority (HKMA): For its pilot exercise, HKMA has consid-
ered the NGFS reference scenarios, as well the RCP scenarios by the IPCC. Transi-
tion scenarios will capture both a disorderly transition and an orderly transition. The
physical risk scenario will focus on the projected climate situation of Hong Kong
in the 21st century, including temperature increase, rise in sea levels and severe
cyclones (HKMA, 2020).
◾ Monetary Authority of Singapore (MAS): MAS has stated that it “will reference
climate scenarios developed by the NGFS” when conducting its climate stress
test at the end of 2022. The exercise will use a range of physical and transition risk
scenarios (MAS, 2021).
◾ Autorité de Contrôle Prudentiel et de Résolution (ACPR): NGFS Phase I reference
scenarios were used as a starting point for the climate pilot exercise (ACPR, 2020).
◽ Reference scenario: Reflects the NGFS orderly transition scenario. It corresponds
to the National Low Carbon Strategy (SNBC) for France to comply with its Paris
Climate Change Agreement commitments and reach net-zero emissions in 2050.
◽ Late reaction: Reflects the NGFS disorderly transition scenario. The GHG emis-
sion reduction objective is not reached in 2030 and requires additional measures.
◽ Scenario of swift and abrupt transition: Carbon prices are revised from 2025,
which leads to a productivity shock. It is assumed that renewable energy produc-
tion is not as mature as in the reference scenarios. Higher energy prices and need-
ing greater investments have negative effects on productivity with the economy.
◽ Business as usual: Mitigation efforts are limited. Measures that were taken
between now and 2050 have very little or no impact on the physical risks. It is the
most pessimistic scenario.
◾ Bank of England (BoE): The climate stress test exercise builds upon the NGFS
Phase II scenarios with three distinct scenarios (Figure 21) (Bank of England, 2020).
◽ Early policy action scenario: Early and decisive action to gradually reduce global
emissions to limit a global temperature change to below 2°C. Prices are reallo-
cated in an orderly fashion by financial markets. No macroeconomic shock due
to structural reallocation of the economy to reduce emissions. Physical hazards
increase in frequency but are under control.
◽ Late policy action scenario: A 10-year delay in addressing climate change which
results in an abrupt action to reduce global emissions and limit the global
temperature change to below 2°C. The financial markets hastily reallocate prices
and reprice assets. There is a significant macroeconomic shock that disrupts the
economy. Physical risks increase with high transitional risks.
◽ No additional policy action scenario: No new policies are introduced to combat
climate change. Transition to a low-carbon economy is insufficient to meet
climate goals. The global temperature rises to more than 2°C. Material shocks
occur later in the century. Transition risks are low, but there are extremely high
physical risks. Figure 21 summarises the impacts of the scenarios used by the
BoE in its climate stress test exercise.
No additional
Early action Late action
action
Transition risks ● Medium ● High ● Limited
Transition begins in 2021 2031 n.a.
Only policies that were

Nature of transition Early and orderly Late and disorderly
in place before 2021
Peak UK shadow
1,100
carbon tax (carbon 900
tax and other policies
(2010 US$/tonne
carbon dioside equiv-
30
ilent)
Physical risks ● Limited ● Limited ● High
Mean global warning 3.3

relative to pre-indus-
trial times by the end 1.8 1.8
of scenario (°C)
Mean sea level rise in 0.39

the UK (m)
0.16 0.16
Impact on output Temporarily Sudden contrac- Permenantly

lower growth tion (recession) lower growth
and higher
uncertainty
1.5 1.6 1.5 1.6

1.4 1.4 1.4
1.2
Average annual output
growth in the UK 9per
cent)
0.1
Yr 6–10 Yr 11–15 Yr 26–30 Yr 6–10 Yr 11–15 Yr 26–30 Yr 6–10 Yr 11–15 Yr 26–30
Source: Met Office, Network for Green the Financial System and Bank calculations
Figure 21: Summary of impacts of the CBES Scenarios.
◾ European Central Bank (ECB): The scenarios used in ECB’s internal climate stress
test exercise complement the NGFS Phase I reference scenarios, covering the
orderly, disorderly and hot house world pathways (ECB, 2021, ECB, 2021).
◽ Orderly scenario: Assumes a transition to a low-carbon economy occurs in an
orderly manner to reduce global emissions by 3% annually until 2030 to limit the
global temperature rise to below 2°C. Emission prices gradually increase due to
the implementation of cost-effective policies, which causes companies to change
their business models and consumers to change their consumption habits.
◽ Disorderly scenario: Delayed and sudden implementation of policies to limit the
global temperature rise to below 2°C. Implementation of more stringent policies
from 2030 leads to higher transition risk.
◽ Current policies: No new policies are implemented, which results in a temperature
rise of approximately 3.5°C by 2100. Transition risks remain low. However, a lack
of transition to a low-carbon economy results in increased physical risks, with
climate hazards increasing in frequency and severity.
Exhibit 14: Bank of England: Drawing Climate Scenarios from NGFS

for the Climate Biennial Exploratory Scenario (CBES).
Due to the global accessibility and need for transparency of NGFS scenarios,
the scenarios have been designed to be simple. Therefore, scenario users are
at times required to modify the scenarios to improve their relevance for the insti-
tution. For its climate stress testing exercise, the BoE made some key changes
from the NGFS scenarios, which have been described below.
◾ Global and regional macroeconomic environment
◽ Additional transmission channels that had not been captured by the
NGFS were added to capture all relevant climate risks. Though the scien-
tific basis of these transmission channels may not be as developed, BoE
believes there are good reasons to assume that certain transmission chan-
nels may occur.
◽ Dips in the UK GDP are much steeper in CBES than in NGFS to explore a
severe climate market disruption scenario (Figure 22).
Figure 22: Decrease in GDP in the late action scenario.
◾ Carbon prices and energy mix
◽ BoE included net-zero commitments specific to the UK as the country has
set out additional policies that need to be captured.
◽ For the NGFS scenarios, nuclear energy is completely phased out due to
technological advancements which allow renewable energy to outcompete
nuclear energy. However, CBES needs to consider the geopolitics and long-
term contracts in relation to nuclear energy. As a result, the scenarios were
adapted to include nuclear energy into the UK’s energy mix (Figure 23).
Figure 23: UK primary energy mix for current, early action and late action scenarios.
◾ Physical risks
◽ Localised physical risks needed to be incorporated, therefore additional
physical risks relevant to the UK with enough evidence to support them
were included.
◽ Physical risks tend to underestimate transition risks. To account for the
uncertainty, future risks were included to the present day.
6.1.3 Scenario design and expansion

Scenario analysis for a climate stress test requires four key steps—selection of scenar-
ios to use, variable selection, quantification of risks through modelling and using the
outputs of the analysis for risk assessment. In this section, we provide guidance for
scenarios and variables selection, scenario expansion and modelling approach for
climate stress tests. Figure 24 below highlights these key processes for scenario anal-
ysis in a climate stress test.
Scenario selection Variable selection Quantification and Risk Assessment
and narrative modelling
◾ Identify the purpose ◾ Identify key physical ◾ Select the combina- ◾ Determine the
of this climate and transition risks tion of models to be impact of climate
stress test exercise relevant to the finan- used risks on the insti-
◾ Identify the scenario cial institution ◾ Design quantitative tution by running
horizon ◾ Select suitable phys- pathways of the financial models
◾ Decide the relevant ical and transition scenarios to be ◾ Results can be
sectors and regions risk variables modelled provided in a disag-
for the exercise ◾ Determine macro- ◾ Understand key gregated manner for
economic and assumptions for the the relevant sectors
◾ Determine which
of the reference financial variables scenario pathways ◾ Use of financial
scenarios, e.g. NGFS to be used as inputs and the models metrics to deter-
and IEA scenarios, during modelling being used mine an institution’s
are the most suit- ◾ Design scenario ◾ Identify the inputs resilience to climate
able to use pathways using required for each stress scenarios
reference scenar- model used ◾ Modelling outputs
ios based on the ◾ Determine the key should be both
risks identified for outputs wanted quantitative and
relevant regions and from each model qualitative
sectors used
Figure 24: Key Processes for Scenario Analysis in a Climate Stress Test.
Selecting scenarios for climate stress testing

Reference scenarios are used by regulators when designing climate stress tests.
However, they are also used by financial institutions for internal climate stress testing
exercises.
54% of UNEP FI TCFD’s programme participants prefer to use NGFS scenarios

for their internal climate stress tests. In comparison, 13% of the participants
prefer to use IEA scenarios and IPCC scenarios.
NGFS scenarios are industry recognised, robust and have the capability to be built
around. One of the main reasons provided by financial institutions for choosing NGFS
scenarios as reference scenarios is because they are used by regulators, indicating
regulatory backing as an important factor in scenario selection. The NGFS scenarios
have been designed for the purpose of meeting the needs of regulatory authorities and
central banks in assessing the potential impact of climate risks on the financial sector.
The scenarios are also publicly accessible and are used by a range of authorities globally
which allows for comparability and consistency across different jurisdictions (BIS, 2021).
Below we describe ten recommended steps for scenario selection for an internal
climate stress test. These recommendations have taken into account discussions with
participating institutions from UNEP FI’s TCFD programme, literature reviews and regu-
latory stress tests.
Ten recommended steps for scenario selection for an internal climate
stress test
1. Conduct portfolio analysis from a climate perspective. Institutions should
understand which portfolios they need to stress test and identify sectors of
concern and the geographical distribution of the portfolio.
2. Select a relevant scenario horizon if not provided by the supervisory author-
ity or if performing the test internally.
3. Identify key climate and economic drivers applicable to the institution.
Begin with generally characterising climate risks of concern (e.g. high
temperature, flooding, net-zero policies) and then identify specificities of
these drivers (geographies, jurisdictions etc.)
4. Develop a range of criteria to select reference climate scenarios (e.g. NGFS,
IPCC and the IEA) developed by research institutions that anchor on (or are
closely related to) the drivers/variables identified in Step 3. Criteria should
include regional and sector coverage, types of risks covered by the reference
scenarios, relevant temperature and emissions pathways, applicability and
supervisory support of the scenarios.
5. Determine the combinations of regions and sectors relevant to the institu-
tion for climate stress testing.
6. Consider what assumptions are made in relation to the scenario pathways,
for example, assumptions needed for policy changes, technological develop-
ments and energy use.
7. Where insufficient granularity is offered off-the-shelf, adapt/expand the
reference scenarios geographically and sector-wise.
8. If not already in the scenarios, derive the impact on the identified key driv-
ers to develop a relevant narrative. It is important to ensure that plausibility
limits are applied to the narrative.
9. Assign responsibility to a team to keep up-to-date with the latest develop-
ments in scenario development. This role is suitable for the climate research
and intelligence function of the climate risk team.
10. Implement a process to integrate the latest scenario developments and
emerging methodology standards into the scenario design for climate
stress testing.
Variables for scenario analysis

In a climate stress test, physical and transition risk variables are combined with macro-
economic and financial variables to quantify the impact of climate risks in a scenario.
Physical and transition risk variables are included in climate scenarios which are then
translated to macroeconomic and financial market variables. These variables are then
used to understand the impact of a given scenario on the financial institution (BIS, 2021).
It is important that the variables selected for the exercise link to the scenario narra-
tive (IIF, 2021). Variables for a climate stress test should provide sectoral and regional
coverage for climate scenarios.
A survey for UNEP FI TCFD participants showed that 74% of financial institu-
tions believe it is very important that relevant variables are broken down by
sector, and 63% of financial institutions believe it is very important that relevant
variables are broken down by region.
The greater the climate and financial risks covered, and the greater the number of
portfolios and regions covered in a climate stress test, the wider the range of vari-
ables needed to be specified by the institution. Climate risk variables commonly
include physical hazards, carbon price and GHG emissions. Macroeconomic variables
commonly include GDP, unemployment and inflation (BIS, 2021). Figure 25 from the
BoE provides an indicative list of scenario variables proposed for CBES in 2019 (Bank
of England, 2019).
Climate risk variables Macrofinancial variables

Physical variables Transition variables Macroeconomic Financial markets
variables variables
◾ Global and regional ◾ Carbon price path- ◾ Real GDP (aggre- ◾ Government bond
temperature pathways. gate and decom- yields for major
ways. ◾ Emissions pathways posed by sector). companies.
◾ Frequency and (aggregate, and ◾ Umemployment. ◾ Corporate bon yields
severity of specific decomposed into ◾ Inflation. for major econo-
climate-related world redions and mies (investment
◾ Central bank rates.
perils in regions with sectors). grade and high
material exposure ◾ Corporate profits yield).
◾ Commodity and
(including UK flood, (aggregate and
energy prices ◾ Equity indices.
subsidence and decomposed by
(including renew- ◾ Exchange rates.
freezez). sector).
ables), by fuel type.
◾ Household income. ◾ Bank Rate.
◾ Longevity. ◾ Energy mix.
◾ Agricultural produc- ◾ Residential and
tivity. commerical prop-
erty prices.
Figure 25: Key variables indicative list by the Bank of England.
For a regulatory climate stress test, it is essential for the regulatory authority to spec-
ify the key variables. Financial institutions will need to derive variables that are not
provided but are useful to analyse for their institution. This was also stated by the
BoE in its discussion paper on the 2021 Biennial Exploratory Scenario (BES) exercise,
explaining “The BES would not provide every variable that participating firms would
need to model the scenarios. In line with other Bank stress tests, participating firms
would have to undertake scenario expansion to extrapolate additional scenario vari-
ables needed to estimate impacts on individual counterparties.” (Bank of England,
2019). Regulators, such as the ACPR and PRA, have provided variable pathways for
their exercises at both the regional level and the global level. Variables were designed
to be applicable at the country, regional and sector-level. Physical variables included
sea level rise, changes in storm patterns and temperature rise. Transition pathways
include carbon price and emissions pathways. The main macroeconomic variables
provided by the ACPR included France, US and EU GDP, inflation rate and unemploy-
ment rate (ACPR, 2020).
Variables needed may vary depending on the scenario, as different scenario narra-
tives will require different variable pathways with varying degrees of material impact.
For example, in the update provided by BoE on its approach to the Climate Biennial
Exploratory Scenario, the regulator explained, “In the Early Policy Action scenario the
Bank does not expect the paths of macroeconomic variables to differ much from trend,
so they are not expected to have a material impact. By contrast, structural changes to
the economy related to the transition from reducing emissions will be crucial. The Late
Policy Action scenario will include some macroeconomic disruption, but structural
changes will continue to be important. For both those scenarios, physical damages
will be part of the scenario. The No Additional Policy Action scenario will involve both
macroeconomic changes and physical damages from climate change.” (Bank of
England, 2020).
Based on group discussions with UNEP FI’s programme participants and taking into
consideration the variable requirements of various supervisors, below we provide
suggested variables for a climate stress test.
Table 15: Recommended variables for climate stress testing.
Variable Type Variable

Price ◾ Carbon price
◾ Commodity and energy prices:
◽ Primary energy price (oil, coal, gas)
◽ Secondary energy price (electricity and gas)
◽ Renewable energy price
◾ Government policies and other taxes
Production ◾ Input costs
◾ Total primary energy (EJ)
◾ Total energy consumption
◾ Amount of fossil fuels used
◾ Amount of renewables used
Investment ◾ Energy efficiency
◾ Energy supply
◾ New technology investments
Emissions ◾ GHG emission targets and pathways (region and sector-wise)
◾ Energy mix (coal, gas, oil, renewables)
◾ Total energy consumption
◾ Government policies
◾ Sectoral GHG emissions (e.g. transportation capacity and cropland)
Physical risk ◾ Temperature (global and regional)
◾ Precipitation and flood risk
◾ Drought risk
◾ Other physical hazard risks
◾ Agricultural productivity, crop yields and food security
Macro-financial and ◾ Household income
socio-economic ◾ Residential/commercial property prices
◾ Unemployment rate
◾ GDP
◾ Corporate profits
◾ Bond yields
◾ Inflation rate
◾ Trade flows
◾ Interest rates and exchange rates
◾ Bank rate
◾ Equity indices
◾ Credit growth (Household, Business)
◾ Credit spreads
If a financial institution is conducting an internal climate stress test or has not been
provided with a complete set of useful variables, it will be required to undertake expan-
sion to extrapolate additional variables for the scenarios. Firms can undertake addi-
tional methods to obtain the scenarios with the pathways required for effective analysis,
depending on the scope of the exercise. Below we have provided a set of recommenda-
tions for financial institutions to follow when undergoing scenario expansion.
Nine key recommendations for scenario expansion
1. Select a baseline reference scenario. If conducting an internal climate stress
test, select a reference scenario that will be expanded for a suitable financial
assessment. Typically, a baseline scenario does not assume transition and
physical risks.
2. Characterise the reference scenario, including understanding the narrative
of the scenario and its assumptions, such as its macroeconomic and finan-
cial factors.
3. Conduct a portfolio analysis to understand clients’ assets, business models
and operating locations.
4. Identify potential climate and financial risks that are relevant to the portfolio
based on the portfolio analysis conducted in Step 3. Risk drivers, especially
macro-financial drivers, that are not covered by the reference scenarios will
need to be expanded, in order to be covered in the scenario analysis.
5. Identify relevant scenario variables for performing granular risk assessment
on the selected portfolio. The financial institution should develop a range of
criteria to assess the relevance of various scenario variables to determine the
ones which are the most relevant. The criteria should analyse the coverage
of variables for a company’s revenues and costs. The criteria should also be
used to understand the variables provided by the regulator.
6. Determine the combinations of regions and sectors most suitable to the
institution for climate stress testing. This will help institutions break down
relevant variables by sectors and regions and project important sectors and
regions not captured by the reference scenario.
7. Using the criteria, determine the variables that are needed but missing from
the list of variables provided by regulators if performing a regulatory climate
stress test. Variables should be selected that will provide the expanded
scenario with adequate characterisation to translate the potential impacts of
climate change into financial impacts.
8. Undertake scenario expansion to extrapolate the additional scenario vari-
ables required for the climate stress test. For effective extrapolation, the
institution should determine the relevant assumptions for extrapolating
scenario variables.
9. Leverage the firm’s current processes and infrastructure for scenario expan-
sion. The climate scenario expansion framework should be tailored into
traditional stress testing scenario expansion frameworks as well as the insti-
tution’s computation systems (e.g. pricing engines).
6.2 Models for climate stress testing
6.2.1 Types of models used

Different financial institutions have come up with a range of different approaches to
model climate risks. For example, some institutions are developing in-house models
while others are acquiring models externally. For a climate stress test, financial institu-
tions should adopt current modelling methods used in a traditional stress test to trans-
late climate risks into financial impacts.
Survey responses from UNEP FI’s TCFD programme participants show that only
7% of participants are quite confident using their current models for a climate
stress test.
This was further emphasised by the ECB in their climate stress testing exercise stating:
“Modelling frameworks, therefore, need to

incorporate plausible representations of both the
economy and the climate in a way that clarifies how
they interact, as well as how policies to prevent or
mitigate climate change affect them both over the
long run.”
ECB, 2021
However, modelling methodologies for a climate stress test are still in their infancy.
The modelling process for a climate stress test consists of—modelling climate vari-
ables, measuring the impact of climate risks on macroeconomic variables, breaking
down the macroeconomic impacts per sector or per portfolio and quantifying the
impact on the financial institution (Figure 26). The modelling approach is complex,
creating uncertainty in the analysis as it involves combining different models that were
not designed to work together. Climate scenario models are needed to project path-
ways for the selected physical and transition risk variables. Macroeconomic models
are then required to translate climate variables from the climate model to selected
macroeconomic variables. Macroeconomic models are an important component
of the climate stress test as they help translate climate risks into financial impacts.
Climate stress testing requires the use of models to breakdown the macroeco-
nomic impacts down to the sector-level, for which sectoral and portfolio models are
commonly used. Following this, financial models at an institution-level can be used to
calculate a bank’s exposures to climate risks. Internal models can be used to assess
changes in metrics, such as probabilities of default (PD) and loss given default (LGD).
Table 16 provides descriptions of the commonly used models for climate scenario
analysis.
Physical risk variables Transition risk variables
Climate Scenario Model
(e.g., temperature rise, flood (e.g., government policies,
(IAM models)
risk, drought) technology investment)
Macrofinancial variables
Macroeconomic Model
(e.g., inflation, employment)
Sector-level disaggregation
Sectoral and Portfolio Models
Climate risks for segments of a portfolio
Financial Models Loss Given Default (LGD) Probability of Default (PD)

(Rating models and counterparty effects (e.g., reduction in effects (e.g., higher costs,
specific credit models) property value) unemployment)
Probability weighted Potential impact of climate

Expected Credit Loss (ECL) risks on business model (e.g.,
Risk-weighted assets profitability, capital & liquidity)
Figure 26: Modelling approach for climate stress testing.
Table 16: Commonly used models for climate scenario analysis.
Potential models Description Open-source/

License
required
REMIND-(PIK) Integrated assessment model, combines socio-eco-
nomic, climate and land-use assumptions. Used in
MESSAGE-(IIASA)
NGFS reference scenarios and initial climate stress
Fully open-source
GCAM - (PNNL-UMD) test scenario design.
IMAGE - (PBL) Integrated assessment model, combines socio-eco-

nomic, climate and land-use assumptions.
World Energy Model Granular energy system model, used to set IEA’s
(IEA) world energy outlook forecasts. Forthcoming License required,
net-zero scenario. some assumptions
Energy Technology Bottom-up model covering energy supply and tech- proprietary
Perspectives (IEA) nological development in the energy sector.
NiGEM (NIESR) License required,
General equilibrium model used to produce macro-
some assumptions
economic variables. Now working with NGFS to
proprietary, NGFS
produce macro outputs for their scenario pathways.
data open-source
GTAP (Purdue) General equilibrium model, multiple regions and
Fully open-source
sectors available.
G-Cubed (McKibben) License required,
Macroeconomic model used by IMF for economic
some assumptions
scenarios on mitigating emissions and warming.
proprietary
One Earth Climate Bottom-up model designed to provide specific Some data and
Model (UTS) sectoral decarbonisation pathways for members of assumptions open-
the UN Net-Zero Asset Owners’ Alliance (NZAOA). source
En-ROADS/C-ROADS Dynamic model allowing users to alter numerous Tool open-source,
(MIT) assumptions around technology and economics to data/assumptions
attain particular temperature outcomes. proprietary
AIM (NIES) Assesses the economic impact of policies to reduce
GHG emissions, with a focus on the Asia-Pacific Free-of-charge
region.
Industry models Used by fossil fuel firms to forecast future demand, Data open-source,
(Shell, BP, etc.) now expanded to function as decarbonisation path- some assumptions
way models. proprietary
Climate scenario models

Climate scenario models project pathways for selected physical and transition risk
variables in a climate stress test. Integrated Assessment Models (IAMs) are comput-
er-based models that generate emission pathways that can be integrated into climate
models. These climate scenario models study potential changes to the climate system
and show changes in production and use of energy over time, technological changes,
natural resource use, and impacts of climate policy (NGFS, 2020). The model combines
varying parameters and metrics to model human society alongside parts of the climate
system. Financial institutions have adopted IAMs more recently due to their ability
to describe the interactions between economic activity, GHG emissions and climate
change. IAMs provide global coverage, have long time horizons and give adequate
details on mitigation pathways that provide high-level overviews of tradeoffs and
constraints. The model optimises behavior across costs, performances, and producer
and consumer responses. IAMs do not consider non-monetary preferences, energy
security and potential sociopolitical interactions. As a result, cost-optimal pathways
may be less desirable and less politically viable than non-optimal pathways (UNEP FI,
2021). In their climate stress testing activities, the ACPR and the BoE used IAMs to proj-
ect trajectories for GDP, carbon price and GHG emissions for each scenario. The outputs
from the IAMs were then used as inputs for the macroeconomic models.
Examples of IAMs include GCAM, MESSAGE-GLOBIOM, REMIND-MAgPIE, WITCH, AIM
and IMAGE. Responses from financial institutions participating in the TCFD programme
showed that 67% of the institutions are using the REMIND model for their analysis and
33% of institutions are using the MESSAGE model. None of the participating institutions
are using the other available IAMs for their modelling. The NGFS scenarios have been
developed to run on three IAMs, namely GCAM, MESSAGE-GLOBIOM and REMIND-MAg-
PIE. These models can differ in respect to other assumptions and coverage (BIS, 2021).
Macroeconomic models
Macroeconomic models are required to translate climate variables from the climate
model to selected macroeconomic variables. Macroeconomic models are an import-
ant component of the climate stress test as they help express climate risks as finan-
cial impacts. A common macroeconomic model used for climate stress testing is
the NiGEM model. ACPR, BoE and DNB have used/are using the multi-country model
for their climate stress test exercises. The model uses a range of macroeconomic
variables such as GDP, inflation, unemployment and interest rates, to determine the
economic impacts of climate risks. The dynamic nature of the model allows the projec-
tion of pathways over a long timeframe. A multi-country model like NiGEM allows
financial institutions to assess both their international exposures, but also exposures at
the country level. In their 2018 energy transition risk stress test, the DNB emphasised,
“using a multi-country model allows us to take account of the fact that energy transi-
tion risks can have global impacts. Given the international exposures of financial insti-
tutions in the Netherlands, global simulations are more relevant than simulations from
a model that considers only the Dutch economy.” (DNB, 2018) Countries are linked
through trade prices, domestic prices, exchange rates, assets held and income flows.
According to the ACPR, the NiGEM model was appropriate to use for their climate
stress test exercise because “Although NiGEM is not a climate model, it has benefit-
ted from extensions to simulate macroeconomic scenarios for climate transition anal-
ysis, mostly associated with public policy”. The use of demand variables like consumer
prices and public spending helps the NiGEM model complement sectoral models for
analysis (Banque de France, 2020).
Sectoral and portfolio models

Climate stress testing requires the use of models to breakdown the macroeconomic
impacts to the sector-level. Sectoral models are used by financial institutions to
downscale the aggregate outputs from a macroeconomic model to the sector-level.
The Financial Stability Institute’s report “Stress-testing banks for climate change—a
comparison of practices” (2021) further explains that the “contribution of each sector
to the aggregate becomes the sector’s gross value added.” For example, for the BoE’s
pilot climate stress test exercise, the regulator developed vulnerability factors that
account for physical risks and emission profiles to identify the sectors that were highly
exposed. The vulnerability factors were then used to calculate the gross value-added
paths for each sector. To downscale macroeconomic aggregates to the sector-level for
its climate pilot exercise, ACPR used an in-house multi-country and multi-sector frame-
work (SSRN, 2020) to assess the impact of carbon prices and productivity shocks on
55 sectors classified by the World Input-Output Database (WIOD). For its climate stress
test, DNB developed transition vulnerability factors (TVFs), based on carbon emissions,
for 56 sectors to determine their vulnerability. Using stock prices determined from the
NiGEM model, they determined the impact of transition risks on sectoral stock prices
to calculate potential sectoral losses (BIS, 2021).
Financial models
Financial models can be used to calculate a bank’s exposures to climate risks. These
models can assess changes in metrics, such as probabilities of default (PD), loss given
default (LGD) and exposure at default (EAD), due to climate risks. Firms use PD analy-
sis to calculate expected loss and it is represented as a percentage of the likelihood of
default. LGD represents the share of an asset lost by a firm when a borrower defaults
on a loan. EAD predicts the amount of loss a bank might face when a debtor defaults
on a loan (Investopia, 2020). PD, LGD and EAD models are important for climate stress
testing to translate the impact of climate scenarios into financial variables for the insti-
tution. Using in-house rating and risk models can help financial institutions determine
the potential impact of climate risks on lending to borrowers and potential losses (IIF,
2021). A bank can adopt the use of internal models to measure the exposure of coun-
terparties to climate risk and incorporate their internal ratings-based approach, to
measure counterparty credit risk and EAD by financial institutions to meet regulatory
requirements (BIS, 2020).
Exhibit 15: Modelling strategy for the ACPR and Banque de France
climate stress test exercise.
In July 2020, Banque de France (BdF) released a proposed analytical framework
for its pilot exercise. The supervisor combined climate models, a macroeco-
nomic model, a sectoral model, a rating model and financial modules for the
exercise.
The modelling strategy began with inputs from IAMs which were used to obtain
trajectories of the climate scenarios used. The models generated trajectories
for GDP, carbon prices and GHG emissions for several countries, including coun-
try blocks (France, EU, USA and the rest of the world). Carbon price trajectories
were then used as inputs for the multi-country macroeconomic model, NiGEM.
Outputs from IAMs were used to set the carbon tax rate in the NiGeM and
sectoral models. To show how these models link together, BdF stated, “for the
baseline and the delayed transition scenario (Scenario 1), productivity shocks
are calibrated in both NiGEM and the sectoral model so that the combined
impact of carbon tax and productivity shocks matches the GDP trajectories
given by the IAMs, ensuring consistency across the three models. For the
sudden transition scenario (Scenario 2), we depart from the GDP trajectory
implied by the IAMs: we assume no productivity gains in the sectoral model,
and let GDP adjust endogenously to the carbon tax shocks.”
The NiGEM model provided macroeconomic and financial variables for the
global economy. The in-house sectoral model is used to translate variables from
the scenarios into potential impacts at the sector-level as value-added. Results
from the NiGEM model are coupled with the sector model to provide a disaggre-
gated output for France, EU, USA and the rest of the world.
BdF then ran the outputs from its sectoral model through a ratings model to
assess the financial impact of climate risks at the firm-level. The further disag-
gregation of results helps determine which firms are the most vulnerable to
climate risks and pose the greatest credit risk for the financial institution. The
ratings model was used to assess credit risk for France. The regulator also
developed financial modules in order to “translate the macroeconomic, sectoral
and firm-level projections into financial variables, such as yield curves, asset
prices and interest rate spreads of corporate bonds.” For example, a dividend
discount model was used to estimate asset prices for sectors and regions and
corporate credit spreads were calculated for sectors and economic areas.
Figure 27 below shows the modelling framework developed by BdF/ACPR for
their pilot exercise.
Figure 27: The modelling framework developed by BdF/ACPR for the

pilot exercise.
(Banque de France, 2020)
6.2.2 Challenges of using existing models for climate
stress testing
Models required for a climate stress test were not initially designed for this purpose.
As a result, financial institutions face numerous obstacles when using models for the
exercise. Below we have identified some key challenges when using existing models.
Lack of granularity: For modelling, IAMs depend on simplified assumptions. As a result,
the output of the analysis might lack the granularity and details needed by risk manag-
ers for decision-making. The long time horizon of the model, as well as its global cover
also limits the depth of the details provided by the model (UNEP FI, 2021).
Uncertainty: UNEP FI’s Pathways to Paris report concluded that simplified assump-
tions about the real world and a long time horizon create uncertainty in the outputs
of IAMs. Furthermore, IAMs are complex models that require numerous assumptions.
As a result, it is very difficult to determine which assumptions have led to a given IAM
output, further adding uncertainty to modelling (UNEP FI, 2021).
Design limitations: The Climate Financial Risk Forum (CFRF) has highlighted the
limitations of economic modelling of climate impacts for high temperature warming
scenarios, such as those for warming at 3–6°C, explaining, “economic models tend to
have mild or simplistic damage functions that fail to respond in a way that is consistent
with the scientific analyses and expectations.” For example, many economic models
project net economic growth at 5°C warming or greater (FCA, 2020). Similarly, in its
energy transition risk stress test exercise, DNB highlighted the limitation of the NiGEM
model, like other macroeconomic models, explaining, “they are not really designed to
simulate the type of structural economic shifts that may follow from the transition to
a low-carbon economy”. Macroeconomic models typically assume that economic rela-
tionships remain stable over time, using historical data (DNB, 2018).
Historical data: Models used for climate stress testing are often calibrated using histor-
ical data or statistical relationships (BIS, 2021). The use of historical data and statistical
relationships can reduce the accuracy of a climate risk assessment as the models may
not be able to capture the full impact of potential climate risks. For an accurate assess-
ment of climate risks, models need to use forward-looking data (IIF, 2021).
Lack of understanding: Due to a lack of understanding of the relationship between
climate risks and macro-financial variables, financial institutions find it difficult to cali-
brate and validate models (IIF, 2021). Greater efforts need to be directed towards link-
ing climate science and financial modelling (BIS, 2021).
Considering the challenges faced by financial institutions when using existing models,
below we provide recommendations for climate stress test modelling.
Key recommendations for modelling approaches in a climate
stress test
1. Determine the parameters required for modelling. The financial institution
should decide the countries that need to be modeled individually, the regions
that need to be modeled in aggregate and the sectors that will be modeled.
2. Define the objective, inputs and outputs and the required components for
each model in the modelling approach.
3. Develop a range of criteria to determine the types of models to use.
4. Survey existing models and providers and determine which to use using the
criteria developed. Financial institutions should use criteria to assist them in
deciding which models are the most appropriate for them to use in translat-
ing climate risks into financial impacts at the firm-level.
5. Identify gaps in the current in-house models using the criteria. Institutions
should be able to identify areas in which their current modelling architecture
can be adapted to incorporate climate variables and identify credit risks. For
example, internal credit rating models can be adapted to identify credit risks
from climate change and perform analysis at the counterparty-level.
6. Increase communication with external modelers and academics for
scenario expansion and to improve understanding of different models.
Increase expertise on modelling in-house through initiating dialogue with
industry experts.
7. Improve collaboration with climate local modelers and research institutions
to develop internal models and improve applicability of internal models to
external models needed for climate stress testing.
8. Further develop in-house modelling capabilities with the guidance and
specifications for models provided by supervisory authorities.
7. Outputs and
applications of climate
stress testing
Key messages
◾ Supervisory climate stress tests have a wide range of applications, includ-
ing being a learning exercise to mobilise and raise the firm’s awareness of
climate risks, encouraging boards to understand the challenges and take a
strategic approach to managing risks, improving a firm’s climate risk manage-
ment, modelling and data, and improving a firm’s climate-related disclosures.
◾ Firms need to ensure clear, consistent and wide-ranging communication and
ensure climate stress tests are a key opportunity to learn and trigger action.
◾ A climate stress test approach should initially be kept simple and focused on
material exposures.
◾ Results can be further leveraged to improve a firm’s climate risk manage-
ment and help set the risk appetite, support the firm’s strategy, customer
engagement and investment, and support the firm’s external disclosures
and compliance.
7.1 Expected outputs of regulatory climate

stress tests
As shown in Table 3 in section 3.3.1, the climate stress test exercises initiated by
central banks and regulators are not identical and exhibit common features in terms
of quantitative and qualitative outputs. Whether it is a regulator-mandated or internal
exercise, the identification of vulnerable sectors in the institution’s global portfolio is
one of the key outputs.
Quantitative outputs: For the banking book, the regulators have been looking to ascer-
tain the impact of scenarios on the expected credit losses, probabilities of default (PD)
and loss given default (LGD) and the stock (or flow) of credit provisions at different
points in the scenarios. The BoE exercise added the requirement to account for the
mitigation and adaptation plans of counterparties if they are underway and highly
likely to be completed. Participants are also required in some exercises to submit their
projected Risk-Weighted Assets (RWAs) and Common Equity Tier 1 (CET1) ratio.
Where the trading book is included in the exercise, regulators have requested fair value
revaluations based on the instantaneous application of the scenarios’ market shocks
and their analysis by sector. This expectation will often be a challenge for market risk
systems which are largely based on individual risk factors and portfolio views, and thus
require substantial manual adjustments and aggregations. The impact of the scenario
shocks on counterparty credit risks may also be included.
Qualitative outputs: All regulatory and internal exercises to-date have also included
a qualitative component to provide an overarching narrative around the participants’
results and methodological choices, to describe in detail the management actions they
would anticipate taking under each scenario, and to capture how business models
could change and how sustainable they are. Qualitative questionnaires are used to let
participants provide their qualitative views of the climate-related risks and opportu-
nities they have identified and their progress towards climate-focused and alignment
metrics. A set of questions may be dedicated to obtaining the participants’ present
(planned) and future (anticipated) approaches to managing climate risk. For example,
for its 2021 Climate Biennial Exploratory Scenario (CBES), the BoE and the PRA have
designed a questionnaire for participants to capture climate risks for financial institu-
tions and to understand current and future plans of participants for managing these
risks (Bank of England, 2021).
Exhibit 16 offers a summary of the key outputs of the ACPR 2020 Climate Pilot Exercise.
Many factors and variables can influence the outputs of climate stress tests but some
of the most impactful drivers are the assumptions around having a static or dynamic
balance sheet. By allowing, or disabling, the impact of the institutions’ strategies, poli-
cies and management actions, these assumptions may result in significant changes in
terms of balance sheet structure and credit exposures to climate risk.
Exhibit 16: Qualitative and quantitative outputs from ACPR 2020
climate pilot exercise.
For reference, a schematic representation of the transition and physical risk
scenarios included in the ACPR exercise is given below.
Figure 28: A schematic representation of the transition and physical risk

scenarios included in the ACPR exercise.
◾ Based on the exercise, the ACPR concluded a “moderate” exposure of the

involved French banks and insurers to climate risks. The results stated that
for the tested institutions 50% of the exposures were in France and 75% were
in Europe, which are less affected than exposures in other regions such as
the United States, which appeared more sensitive to transition risk.
◾ As identified in this exercise, the exposure of French financial institutions to
sectors most impacted by transition risks is relatively low. Under a dynamic
balance sheet assumption, the sectoral structure of the credit exposures
for the six banking groups shows a distortion under the sudden transition
scenario, as shown in Figure 28, to the detriment of the sectors negatively
affected by the transition scenarios.
Figure 29: Evolution of the sectoral structure of credit exposures under the
sudden transition scenario.
◾ The dynamics of the results in Figures 29 and 30 show the significant

impact associated with an increase in the price of carbon and a slow-down
in GDP growth.
Figure 30: Evolution of the aggregated annual cost of risk for

participating banks.
Figure 31: Weighted average of the probabilities of default (PD) by sector over time.
◾ The exercise determined that participating institutions are at risk from vari-
ous physical hazards, including droughts, floods and cyclones; yet the prog-
ress made in taking physical risk into consideration is very limited, with only
two institutions able to quantify the impact of an increase in the lack of insur-
ance coverage on its credit risk parameters. The report linked this result to
the difficulty of obtaining a precise geographical location of their exposures
at a consolidated level.
◾ Banks and insurers need to step-up their efforts to integrate climate risks
into their current risk management processes, taking the exercise results into
account and promoting a better allocation of their resources.
◾ The outputs of the exercise also helped identify methodological limitations
in need of improvements, including improving the models and data sources
used, better integration of physical risks into the analysis and improving the
hypotheses used for creating scenarios and identifying at-risk sectors.
(ACPR, 2021)
7.2 Applications and uses of climate stress tests

outputs for supervisors
In a July 2021 report (BIS, 2021), the Financial Stability Institute at the Bank for Inter-
national Settlements discussed the challenges that emerge when trying to adapt tradi-
tional stress tests to banks’ climate-related risks. Using recent stress tests exercises
(DNB, BdF/ACPR and BoE/PRA), the report reviewed how these challenges have been
addressed in practice and provided some reflections about the possible implications
for prudential requirements. Exhibit 17 summarises the key outcomes of climate stress
testing identified in this report. We further develop the key applications of climate
stress tests below.
Exhibit 17: Key takeaways on climate stress testing outcomes

from the BIS report ‘Stress-testing banks for Climate Change—a
Comparison of Practices’.
◾ A climate stress testing exercise provides an initial assessment of climate
risks, exposures and vulnerabilities to which institutions are exposed. The
report states that “these initial exercises are primarily a learning opportunity
for banks and the authorities”.
◾ Quantification of exposures to climate risks can “guide banks in rebalancing
their exposures, and/or adjust their risk management accordingly” and can
help authorities to “gauge the financial stability implications of climate risk”.
◾ Outcomes of a climate stress test can measure the significance of antic-
ipated impacts, illustrate sensitive regions and sectors, and help assess
corrective actions.
◾ Results of climate stress tests are not currently being used to set capi-
tal requirements for banks. The exercises which have taken place to-date
have not quantified solvency ratios at the bank level and have reported their
results at the aggregate level.
◾ Regulatory authorities are not disclosing the results publicly at the firm-level
due to high uncertainty surrounding the produced estimates. These can
nevertheless be useful on a system-wide level and, for banks, as a way to
benchmark their own calculations.
(BIS, 2021)
A learning exercise to mobilise and raise institutions’ awareness of

climate-related risks
With expertise in modelling climate-related risks still in its infancy, regulators anticipate
or have reported significant benefits of such exercises in developing the capabilities and
resources of both the regulator and participating banks. The ACPR (2021) stated that:
“…participating institutions (in the ACPR pilot climate stress test)

commended the benefits of this pilot exercise and the progress it
has fostered in terms of cross-functional mobilisation of teams,
internal reflections on risk analysis and the limits of the models
currently used, but also in terms of strategic guidelines and towards
a better understanding of the issues and the impact of climate
change on their business model.”
Climate stress tests provide an initial identification and assessment of the climate-related
risks and vulnerabilities that institutions could be exposed to and the potential costs of
their non-compliance with the objectives of the Paris Climate Change Agreement.
A way to encourage boards to understand the challenges and take a

strategic, long-term approach to managing climate risks
Climate stress tests will shed light on the financial risks arising from climate change
and focus the attention of institutions on impacts including potential losses or degra-
dation of the credit quality of exposures. Climate stress tests can also create the
impetus for boards to understand the challenges that climate risks represent for their
business models and consider strategy changes and other management actions that
might be taken, as well as define the likely triggers and thresholds for their implemen-
tation under each scenario.
With the knowledge of such anticipated corrective actions, regulators may be able to
identify some of the expected tensions, as highlighted by ACPR (2021), for example
between exit strategies and market share retention objectives or the desire to maintain
a customer relationship. These may lead to the retention of an exposure to transition or
physical risks for a longer period of time than anticipated.
A contribution to improving institutions’ climate risk management,

modelling and data
Climate stress tests allow banks and regulators to identify data and resource gaps, any
shortcomings in existing measurement tools, as well as the appropriate indicators and
metrics to measure and monitor the impact of climate risks on the financial sector.
In doing so, regulators help improve the banks’ abilities to anticipate and manage these
risks by developing each institution’s capabilities to integrate the climate risk in their
risk measurement, assessment and management tools and frameworks. For example,
in the results for its 2018 energy transition risk stress test, DNB (2018) concluded: “The
stress test results suggest that financial institutions can mitigate their vulnerability to
a disruptive energy transition by including energy transition risks in their risk manage-
ment.” Although not explicitly stated, it is likely that regulators will use the results of
climate stress tests to benchmark banks within their remit, using the best approaches
to set the expected standard, the level of which will increase as banks’ approaches
develop and mature.
A contribution to improving banks’ climate-related disclosures

A key benefit is also that such climate stress testing exercises will encourage banks to
improve their climate-related disclosures on a comparable basis which will increase
transparency on the risks and opportunities for the financial sector. Improved disclo-
sure will also inform investors and market actors in general on the resilience and
sustainability of institutions, progress towards targets and their alignment with the
goals of the Paris Climate Change Agreement.
For now, a supervisory tool with no direct regulatory capital implications
As argued in a 2020 research note by the Bank Policy Institute (BPI, 2020) the challenges
in stress testing for climate change are significant and it thus seems premature “to link
climate stress tests to capital requirements” despite the fact that “the underlying analy-
sis and research will gather valuable information for public policy on climate change”.
As part of all such exercises to-date, the regulators have insisted on the absence of
direct implications in terms of regulatory capital or remedial actions to be taken by
banks as a result of projected losses. In its report, the Financial Stability Institute does
not expect this initial round of climate stress exercises to lead to new capital require-
ments, explaining that “such a requirement is considered premature given the prelimi-
nary nature of the exercises and the high-level of uncertainty attached to their results”.
Yet, regulators see the exams as a prudential supervision tool: in addition to the goals
described above, stress tests also help regulators benchmark firms, assess to what
extent the current regulatory framework is sufficient to manage climate risks and
whether regulatory capital is the most appropriate tool to address them. For example,
the BoE/PRA stated that the 2021 CBES “may inform the Financial Policy Committee’s
approach to system-wide policy issues, the PRA’s approach to supervisory policy, and
guide further work between participants and supervisors to address any issues high-
lighted” for their business model, internal governance and risk management.
Exhibit 18: Desired outcomes of the 2021 CBES by the BoE and
the PRA.
The objective of the 2021 CBES exercise is to understand how resilient current
financial institutions and the financial system are to climate risks. The BoE
released the key elements of its 2021 CBES exercise, in which the supervisory
authority has outlined its desired outcomes, which are:
◾ Sizing the financial exposures of participating financial institutions and the
financial system to climate risks;
◾ Understanding the challenges that the business models of participating
financial institutions will face from climate risks, determining likely responses
and understanding the potential implications; and
◾ Assisting participating financial institutions in enhancing climate risk
management, including the engagement of counterparties to support them in
understanding their vulnerability to climate risks.
These desired outcomes have also been illustrated in the Figure 32 below.
Figure 32: Desired outcomes of CBES.
(Bank of England, 2019)
Whilst it may occur at different times in different jurisdictions, it is certain that climate
risks will be included in all relevant supervisory requirements, as any other risk, at some
point in the future. The ECB has made its expectations clear and asked the banks
within its remit to assess themselves against climate risks, with a view to benchmark-
ing the banks and developing the ECB’s approach to a “dedicated Supervisory Review
and Evaluation Process (SREP) methodology that will eventually influence banks’ Pillar
2 requirements” (ECB, 2021). In June 2021, the EBA published its report on manage-
ment and supervision of ESG risks for credit institutions and investment firms. The
report provides “a comprehensive proposal on how ESG factors and ESG risks should
be included in the regulatory and supervisory framework for credit institutions and
investment firms”. It proposes an enhancement of the SREP, with an extension of the
time horizon of the supervisory assessment of the resilience of institutions’ business
models, “applying at least a 10-year horizon to capture physical risks, relevant public
policies or broader transition trends” (EBA, 2021).
The PRA also made clear since 2019 that they expect banks in its remit to include
climate risks in their ICAAP and thus consider the capital implications of the associ-
ated exposures and risks. They also mentioned that the CBES exercise will “guide
further work between participants and supervisors to address any issues highlighted”
(Bank of England, 2021).
Although climate risks are not accounted for directly through capital requirements for
now, these processes might result in an indirect capital impact. Banks across the EU
and the UK therefore have a strong incentive to identify and quantify their exposures to
climate risks and to understand how well their business model would withstand signifi-
cant climate-related stress through their ICAAP and Internal Liquidity Adequacy Assess-
ment Process (ILAAP), as well as their recovery and resolution planning processes.
7.3 Recommendations on the use of climate

stress tests outputs for financial institutions
Currently, only a small number of climate stress testing exercises have taken place, like
the ACPR pilot exercise, or are ongoing, like the BoE CBES exercise. Though climate
stress testing exercises are becoming common across the financial sector, with more
exercises anticipated to take place over 2021 and 2022 across several jurisdictions
(see section 3.3.1 for the list of regulatory stress tests announced), climate stress test-
ing is still in its early days. Banks need to develop their capabilities, including internal
climate stress tests, to anticipate how they will use their outputs to integrate this think-
ing into their climate programme.
In the first part of this section, high-level guidance regarding the climate stress test
process will help frame the thinking around its design and how to approach it. The
second part focuses on how to best use the climate stress tests outputs and provides
guidance and recommendations on six key areas.
7.3.1 Overarching recommendations for the climate stress

test process
1. Ensure clear, consistent and wide-ranging communication
UNEP FI TCFD programme participants stressed the importance and difficulty of inter-
nal communication: if traditional stress tests can already be challenging to communi-
cate about, climate stress tests, because of the time horizon involved, present another
level of challenge.
Climate stress testing exercises need to be effectively communicated across the firm
and at all levels of management. The board and executive management need to be
clear about what the climate stress tests can and cannot do, what the outputs mean
and what outputs can and cannot be relied upon. To improve communication on
climate risks, HKMA recommends that senior management should both actively take
part in international conferences and initiatives on climate risks to demonstrate their
institution’s commitment to tackling climate change. HKMA also recommends that
firms enhance internal communication to raise awareness of climate risks across their
teams through a range of practices, including circulating articles on climate change
and its impacts (HKMA, 2021).
As described in section 4 (Team Organisation and Skills), management at the firm
plays an important role in effectively communicating the results, as well as the limita-
tions and challenges of the exercise. Regulators also stress (for example in the 2018
European Banking Authority (EBA) ‘Guidelines on institutions’ stress testing’) that the
management body should ensure that the outputs of a climate stress test are taken
into account for decision-making, business planning and strategy (EBA, 2018). BCBS
recommends that senior management clearly discloses the use of climate stress
outputs as a guide when making decisions, taking into consideration the vulnerabilities
shown by the exercise (BIS, 2009).
2. Ensure climate stress tests are a key opportunity to learn and trigger action
Though climate stress testing is in its early stages, there are useful lessons to be
learned from such exercises. Qualitative and exploratory views can greatly benefit
firms by triggering action, for example they can prompt early discussions on climate
risks and opportunities, and they can help devise and test strategies to address poten-
tially problematic exposures. As institutional approaches to climate stress testing
mature and become more sophisticated, climate risks need to be increasingly embed-
ded into decision-making, risk management, business models and strategy. These
improvements should be reflected in disclosures.
A survey of the UNEP FI TCFD programme participants confirmed the regulatory views
discussed in the previous section: banks anticipate that the key applications of climate
stress test results will primarily be to further their understanding of the impacts of
climate risks on the bank’s portfolios (24%), to inform their risk appetite (20%) and risk
management policies (19%).
24%
20%
19%
17%
8%
7%
5%
Informing your Informing your Changing your Informing your Informing Revising your Further develop-
firm’s business firm’s approach firm’s client firm’s product your firm’s risk firm’s risk ing your firm’s
plans to capital plan- onboarding or pricing appetite management understanding of
ning annual review policies climate risks
process
Figure 33: Anticipated primary uses of the climate stress test results by the UNEP FI
TCFD programme participants.
This aligns well with the benefits regulators anticipate: raising awareness of the board
and executive management of the potential impacts, providing an understanding of
climate risk drivers and initiating their inclusion in risk management frameworks and
processes. Importantly, running climate stress tests also allows institutions to recog-
nise how much progress needs to be made on data, infrastructure and capabilities, and
provides incentive to devise plans to improve these.
3. Initially keep the approach simple and focus on material exposures
Unless required otherwise, institutions that have some experience with climate stress
tests tend to adopt a risk-based approach by focusing on the most carbon-intensive
sectors in their portfolios. This is a reasonable approach when resources are limited,
and a significant part of the stress testing process relies on trial-and-error and manual
adjustment processes. These banks have kept things simple, looked to avoid any
undesirable extension of the scope, been clear on their modelling approach and the
assumptions involved, and have pushed increased levels of sophistication to a later
date. They have gone through a lessons-learned process to help drive both investment
and improvements in their capabilities.
4. Look for opportunities to leverage the outputs
The responses from UNEP FI TCFD programme participants also reflect that, while
some banks are more advanced than others, much remains to be done to reach a level
of maturity that would allow them to fully leverage climate risk assessments and anal-
yses to drive their strategies.
The more advanced banks are already looking into how climate stress tests can
support client engagement, identify winners and losers within material sectors under
the assumptions set out by the scenarios and thus inform client strategy. Some institu-
tions are also developing their thinking as to how the outputs could usefully inform risk
appetite, be used to set constraints (for example limits on certain sectoral exposures)
and ultimately reshape capital allocation. These points will be specifically addressed in
the next section and banks are encouraged to leverage the analyses derived from the
outputs to improve the sustainability and performance of their business.
7.3.2 Leveraging climate stress tests outputs

The most immediate output of the analysis and interpretation of the climate stress test
results is the set of insights derived on the impact of physical and transition risks at
different points in the future, along determined pathways. These insights allow finan-
cial institutions to identify which sectors and counterparties are most at risk in portfo-
lios under the scenario assumptions and identify the climate drivers for these risks.
Beyond the immediate benefit of identifying and quantifying such risks, both regula-
tory and ad-hoc climate stress test results present a range of potential benefits which
banks should fully take advantage of. This section provides guidance on how these
exercises and their outputs could be developed, focusing on the six points schemati-
cally represented in Figure 34.
Better understanding of Provide direction for the
vulnerabilities to incorporate into institution’s climate data and
the risk appetite infrastructure investments
Provide value-adding
Inform proposed shifts in management information
business strategy and informative external
reporting and disclosures
Inform customer engagement, Develop mitigation strategies

especially in vulnerable sectors and push the analysis further
Figure 34: Six key uses of a climate stress test.
These intrinsically linked uses can be grouped around three themes:

1. Improving climate risk management and helping set the firm’s risk appetite;
2. Supporting the firm’s strategy, customer engagement and investment; and
3. Supporting the firm’s external disclosures and compliance.
Theme 1. Improving climate risk management and helping set the firm’s
risk appetite
Developing climate stress testing capabilities will allow institutions to identify and
understand their vulnerabilities, develop appropriate risk management approaches,
and integrate climate risks in their risk appetite. While the outputs of climate stress
tests allow institutions to build value-adding management information and dashboards,
they can also lead to improved external reporting and disclosure. Climate stress tests
hold the potential for further uses in probing the tails of the risk distributions and gain-
ing greater understanding of the impacts different pathways can lead to.
Key use 1. Better understanding of vulnerabilities to incorporate into the

risk appetite
The identification and understanding of business vulnerabilities ultimately aims to
inform risk appetites and helps set targets and limits. Climate stress tests allow insti-
tutions to understand what may lie ahead for each sector and material counterparty or
country exposure, how the results vary with the portfolio composition over time, and
how sensitive they are to changes in the data, modelling or assumptions. Understanding
vulnerabilities and how they translate in terms of financial parameters for institutions
and their clients will expand and deepen with experience in running climate stress tests.
Recommendation. Beware of unwarranted precision. Climate scenario analy-
sis and stress testing can be used to estimate the sensitivity of firms’ port-
folios to specific drivers and understand their potential non-linear effects. In
conducting such analyses, relative and order-of-magnitude assessments can
yield useful insights. Institutions should not attempt to claim spurious preci-
sion in their outputs.
As institutions and practitioners have grown accustomed to seeing in the literature, there
are an infinite number of shapes that the future could take, of paths our climate and
economies could follow, and no one can predict which will effectively unfold. Climate
scenario analysis and stress testing hold tremendous potential in that they allow institu-
tions to simulate distinct realisations of the future and—crucially—the path taken to get
there. This in turn can be used to determine their potential impacts on the firms.
Recommendation. Do not limit your analysis to regulatory scenarios. While

regulatory scenarios are a good starting point, each firm should work to estab-
lish its own set of climate scenarios, increasing the level of detail in the sectors
and geographies most relevant to its portfolio and ensuring a relevant variety of
pathways for the key drivers to be explored.
Once climate vulnerabilities, drivers of risk and sensitivities have been identified, anal-
ysed and quantified through stress tests, and the pathways relevant to each scenario
have been understood, they are expected to inform the choice of which sectors to
include in the risk appetite, which controls to set up, which metrics to use and which
constraints or trigger levels to apply. This will be an evolving process in which expert
and qualitative judgment is likely to be initially predominant over modelling and quan-
titative assessments. In addition, the uncertainty around climate stress test results is
expected to represent a challenge for embedding climate metrics in firms’ risk appetite.
Whether their risk appetite has been articulated or is still a work in progress, banks may
wish to implement some form of limitation on exposures to sectors or counterparties
which do not align with their climate strategy or their risk appetite. Ideally, such limits
translate directly to the risk appetite statement or contribute to it in a straightforward
and unequivocal way. Nevertheless, it may prove challenging to translate a qualitative
risk appetite statement into metrics and quantitative limits.
Recommendation. Use metrics derived from the key drivers to define your risk
appetite and leverage existing guidance on climate metrics.
The TCFD released its guidance on the use of forward-looking metrics for the
industry in October 2021. The guidance report provides information on selecting
and disclosing metrics, as well as the set of metrics that the Task Force consid-
ers important to disclose (TCFD, 2021).
The UK Climate Financial Risk Forum also published guidance on forward-look-
ing metrics in October 2021 (FCA, 2021). The guidance identifies key climate-re-
lated metrics and their primary use cases, as well as guidance on developing
metrics to better understand system-wide risks.
The March 2021 EBA consultation on Pillar 3 ESG disclosures (EBA, 2021) intro-
duced the Green Asset Ratio (GAR) in addition to Key Performance Indicators
(KPIs) and quantitative disclosures on climate-change transition and physical
risk. The GAR provides information on what part of an institution’s exposures
“contribute to or enable the objectives of climate change mitigation and adapta-
tion and help to mitigate climate change related risks”.
Some institutions may prefer to start their journey with qualitative or relative
constraints which can meaningfully influence the evolution of a portfolio. For exam-
ple, limits could be set on the percentage of lending to counterparties whose opera-
tions are based in locations subject to increased probability of floods or wildfires or
are classified as carbon-intensive. A portfolio could be constrained to have at least a
certain percentage allocated to green or sustainable assets. Such constraints need not
be based on an accurate quantitative assessment but could rely on a high/medium/
low classification. They may also not be directly linked to climate-derived metrics but
reflect a policy, such as imposing a shorter maximum maturity or a lower loan-to-value
requirement for undesirable products or counterparties.
An institution’s understanding of its climate-related risks and opportunities would not
be complete without considering all possible sources. Regulatory climate stress tests
have so far heavily focused on credit risk as the key risk for institutions, with the BdF/
ACPR exercise also including market risk. As shown in Figure 35, climate risks lead to
many financial risks and all material risks need to be considered in climate stress tests.
Figure 35: Transmission channels for climate risks into financial risks (NGFS, 2021).
Market risk. Climate events may trigger extreme market movements such as (but not
only) volatility in commodity prices. Sudden policy changes, such as the imposition of
a carbon tax or technological changes, may cause the equity prices of companies in
carbon-intensive sectors to fluctuate. As climate events continue to occur with increas-
ing frequency and severity, market risk will be impacted by climate change.
Liquidity risk. Banks suffering climate-related losses may face difficulties in refinanc-
ing and could face a liquidity run due to customer needs for cash in light of climate
risks. Therefore, climate risks have the potential to cause net cash outflows and
depletion of liquidity buffer for banks (ECB, 2020). Furthermore, as credit rating agen-
cies focus increasingly on ESG risks, poor management of climate-related risks could
adversely impact the bank’s credit rating and consequently affect its ability to obtain
liquidity from the market.
Operational risk. Climate change has had a clear impact on operational risk, as
extreme weather may force office closures or damage crucial physical infrastructure
such as buildings, data centres, call centres or bank branches, and impact critical
third-parties. New climate-related technologies also need to be evaluated for the opera-
tional risk they carry for the organisation.
In addition to the above financial risks, reputational risk and legal risk should be consid-
ered, especially in areas where the number of litigations based on climate consider-
ations is increasing or the potential for greenwashing is present.
Recommendation. Early in the process, consider the requirements and impli-
cations of including all material risks in the climate stress test framework and
risk appetite. Climate risk is a “transverse” risk that will manifest itself through
existing risk channels. Credit risk is the main one in the banking industry, but
other risks could potentially inflict serious losses on an institution and need to
be included in the climate risk assessments.
Key use 2. Develop mitigation strategies and push the analysis further
Once a picture of the size and extent of its climate risk exposure starts to emerge from
risk assessments, climate scenarios and stress tests, an institution will look to develop
risk management and mitigation strategies and take advantage of the opportunities.
This can take many shapes, from a planned reduction in trades that adversely contrib-
ute to climate risk in a given sector, to active hedging of climate-induced counterparty
risk, to the development of green financing products, as addressed in the below section
about strategy and customer engagement.
Institutions can also look to further embed stresses in their day-to-day risk manage-
ment practices, including defining some benchmark scenarios that run regularly and are
monitored for progress against a target or risk indicators. For example, developing a set
of specific scenarios aimed at key climate-related exposures or vulnerabilities for given
sectors or industries, or highlighting the specificities of a given region or country, could
be valuable tools in monitoring how an institution’s risk profile evolves. In a similar way,
acute physical risk scenarios depicting plausible, yet catastrophic hurricanes, floods
or wildfires should find their way into every institution’s scenario arsenal. Finally, public
policy-linked scenarios would allow institutions to simulate the impact of the implemen-
tation of one (or several) key policies, simulating different timings, extent or intensity.
Finally, the Reverse Stress Test approach (Exhibit 19) could prove a valuable tool in
identifying the tail events likely to cause serious issues to institutions in the absence of
any relevant historical occurrence.
Exhibit 19: Probing the tails with Reverse Stress Tests.
Reverse Stress Tests (RST) are a hypothetical exercise in which an institution
aims to find scenarios that would lead to a failure of its business model and
cause the market to lose confidence in its ability to function, leading to its
demise. By their very definition, the scenarios generated are extreme and their
plausibility is often questioned. Yet they can be useful to identify vulnerabilities
that may have been overlooked by other stress testing approaches.
Climate change events would qualify well for such an exercise, looking at what a
catastrophic climate event(s) would lead to. While traditional stress tests largely
rely on historical data as their basis, devising stress events of a magnitude
well beyond anything that ever happened is appropriate in the current climate
change context, given the increasing frequency and severity of such events.
Such an approach could allow institutions to answer questions like, what level of
severity would be required for a climate-related event to cause a loss in excess
of a given amount? Or, for a mainly coastal or flood-exposed mortgage portfolio,
what level of severity would cause the portfolio to experience an unsustainable
level of defaults and mostly worthless collateral?
Transition-related climate risks could also be captured in RSTs. During the global
economic recession caused by the COVID-19 pandemic, exposures to troubled
shale oil producers forced US banks to set aside large provisions for non-per-
forming loans. An example of a transition-related RST question could be: what
level of newly imposed carbon tax or level of oil price would cause a pre-deter-
mined proportion of such a portfolio to fall into arrears?
Recommendation. Develop and adopt actionable mitigation plans and adapt

them for the specificities of each sector. The institution should consider ques-
tions of overall strategy, climate commitments and client engagement in consid-
ering how to most effectively support a resilient, low-carbon future.
Recommendation. Develop internal climate stress tests catering to the speci-

ficities of the firm’s portfolio and exposures. Embed transition and physical risk
scenarios into day-to-day risk management, monitoring and reporting processes.
Theme 2. Supporting the firm’s strategy, customer engagement
and investment
Climate stress test outputs will likely identify one or more sectors for which increas-
ing or even maintaining the current level of exposure might lead to adverse outcomes
in one or more scenarios. This should lead to discussions about the current business
strategy, inform any proposed changes, as well as the extent and timeframe of the
decarbonisation of the lending portfolio. These elements will, in turn, drive the engage-
ment of key clients.
Going a step further, by embedding climate stress test outputs into governance, risk
management processes and policies, a firm can make informed, risk-based business
decisions, improve its resilience to climate change and seize the opportunities as
they arise.
Key use 3. Inform proposed shifts in business strategy

While there is an intense focus on the risks associated with it, climate change also
creates opportunities. The TCFD identified several areas of opportunity, including
improvement in resource efficiency, adoption of low-emission energy sources, devel-
opment of new low-emission products and services, access to new markets, and resil-
ience-building along the supply chain. These opportunities create increasing demand
for green funding and require a shift in many institutional strategies.
Shifts in business strategies, actions to tackle the identified sector vulnerabilities and
leverage business development and growth opportunities, should be reviewed in each
institution in the context of its organisational strengths, market position and compet-
itive landscape with the support of climate stress tests. Shifts should also account
for shareholders’ and stakeholders’ expectations, climate-related policies and external
commitments such as science-based targets. Climate stress test results could add
significant value by informing the choice of metrics derived from the key risk driver and
helping to define the risk appetite against the targets supporting the firm’s strategy. A
climate stress test can contribute to strengthening the link between a bank’s current
climate commitments and its desired outcomes, as well as supporting the develop-
ment of clear management outcomes in the case of each scenario to support the
development of the firm’s overall climate strategy.
An appropriate climate stress testing set-up should also allow the firm to identify
opportunities and test the impact of business options. It could include scenarios
focusing on a particular driver, with various options, to help answer the business and
management’s questions and support the decision-making process. For example, what
would be the impact on the firm’s current portfolio if a particular type of government
or regulatory policy was adopted in 2025, 2030, or 2040, assuming the balance sheet
structure does not change? Or how would a proposed internal strategy, for instance
in terms of substantially increasing or decreasing lending to a particular sector—thus
changing the balance sheet structure—play out under the scenarios considered?
As discussed earlier in this section, stress tests should also allow the exploration of
vulnerabilities of the current (or a proposed) business model to future climate path-
ways, for example describing a more or less abrupt reduction of carbon emissions and
associated economic and social repercussions. It will help identify the potential risks
posed to this business model over time under various hypotheses and thus contribute
to building the firm’s resilience.
Often less considered, are the potential impacts of climate change on banks’ liquid-
ity, collateral and capital management. Climate stresses could usefully inform the
management of these areas and risks under various scenarios and support the discus-
sions on strategic planning and sustainability. Regulators have started to set out their
expectations for the ICAAP, ILAAP as well as the recovery and resolution planning
processes (as discussed in section 7.2).
Recommendation. Embed the use of climate stress test outputs in discus-

sions about business strategies across the organisation. Use outputs to
support and challenge business plans, ensure their compatibility with the risk
appetite approved by the board and bring focus to both growth opportunities
and resilience.
Key use 4. Informing client engagement, especially in vulnerable sectors

Financial institutions know their clients’ businesses in detail and are thus in a unique
position to fully understand and anticipate their financing needs, combining this
knowledge with the understanding of climate-related risks obtained from climate risk
assessments and stress tests. Institutions can use climate stress test outputs in their
discussions and engagements with their clients, bringing together net-zero discus-
sions and climate risk discussions. Such engagement will lead to the identification of
business opportunities in the form of new strategies and tailored products and will
also help define the institution’s strategy with respect to clients in vulnerable segments.
As part of their credit assessments, annual reviews and ongoing monitoring processes,
banks should review their clients’ business activities and transition plans against
their own climate strategy or risk appetite. Where they are misaligned, institutions
can adopt one of two approaches, or a mix of both. Firstly, banks can look to put in
place mitigation measures to allow a reduction in exposure over a certain time hori-
zon. Secondly, banks can look to support their client in transitioning to a low(er)-carbon
business model. Mixing the two approaches may, for example, take the form of agree-
ing on specific targets or milestones in their transition plans, providing financing for a
shorter period, introducing stringent covenants on climate-related metrics, or making a
refinancing offer conditional on achieving climate-related targets. In this example, the
climate elements are included at the transaction level for a particular client. This could
prove a very effective approach to both support and encourage the transition of a client
while minimising the risk to the bank.
Climate stress tests can inform these approaches by providing a set of metrics on
which targets can be set, to introduce some limitation on exposures to sectors or
clients that do not align with firms’ climate strategies or risk appetites.
Consistency in defining and applying such rules across the organisation is key and the
sector-specific adopted limitations, for example, should be included in the relevant poli-
cies and procedures, describing the climate-related considerations or criteria to account
for. Any exclusion policy (such as the financing of thermal coal in most EU banks)
should also filter down into credit approval processes, including any gradual measures
to be adopted as relevant. Exhibit 20 explains how climate stress tests can lead to
adjustments in the internal credit rating and thus the pricing of climate risk in firms.
Exhibit 20: Adjustment to credit ratings and pricing of climate risks.

Once a risk has been identified and quantified, an institution would ideally need
to include its impact in the pricing of its products where relevant and material.
Banks should step-up their efforts to use the climate stress test modelling and
outputs to incorporate climate factors into their Internal Ratings-Based (IRB)
credit models and review the pricing of the firm’s current product offering to
ensure climate risks are accounted for, at least to the extent possible.
The uncertainty around the outputs may not yet fully allow this and a first step
towards this goal could be to use the outputs of the stress tests as a qualitative
or semi-quantitative way to inform lending decisions. This would allow changes
to the business strategy, to encourage or discourage lending to certain key
sectors, charging more to counterparties whose business increases the proba-
bility of adverse outcomes for the bank, or does not contribute to the diversifica-
tion of the portfolio in terms of climate risks, for example.
Recommendation. Define how climate stress test outputs will be used in the
firm’s client engagement strategy and ensure easy access to sector-specific
and client-specific analyses.
Key use 5. Provide direction for the institution’s climate data and
infrastructure investments
Undertaking climate stress testing will allow an institution to identify suitable
approaches to transition and physical risk modelling, evaluate the amount of work
required and plan for future improvements.
In the same way, data gaps and data quality challenges will be identified. Section 5
highlighted the challenges to be expected in collecting and processing climate data.
Once candidate solutions, internal or external, have been explored and analysed, they
should be included in specific investment plans and budgets for approval by the execu-
tive management and the board, hence reinforcing their importance.
It is also likely that the early attempts at running climate stress tests will highlight at
least some limitations of the institution’s current systems in incorporating and manip-
ulating climate-related data and metrics. The tools to manage climate stresses and
analyse their outputs are highly likely to be ad-hoc in the early stages and this should
drive the development of a programme of work to address the institution’s deficiencies
in its infrastructure and tools.
Theme 3. Supporting the firm’s external disclosures and compliance

In addition to data representing the extent of key climate exposures, climate stress
tests can help emphasise the quantum of risk, through value-adding management infor-
mation and dashboards. The themes developed above are good candidates to popu-
late these dashboards and allow management and the board to monitor the impact
and progress of actions they are taking to mitigate climate risk, take advantage of new
opportunities and inform business strategy and decision-making on an ongoing basis.
Key use 6. Provide value-adding management information and enhanced

external reporting and disclosures
The use of climate change scenario analysis is a core TCFD recommendation. However,
only the largest banks in the developed markets are currently at the forefront of using
climate scenarios in their disclosures. The situation is improving each year, as disclo-
sures increasingly include more granular and quantitative information and more institu-
tions join the initiative, which is also increasingly being made mandatory by regulators.
As banks mature in their approaches and develop their climate stress testing capabil-
ities, they will increasingly see benefits in leveraging the results of their stress tests
to improve their disclosures. In turn, providing more detailed and insightful data and
analysis will help them meet the expectations of their shareholders and stakeholders
and support their advocacy and communication efforts. It will also facilitate compli-
ance with regulatory disclosure expectations and potentially limit exposure to climate
risk litigation in relation to material omissions or misstatements of climate risks they
are exposed to. The variety of approaches and metrics across banks makes compar-
ative assessments difficult. For example, if an analyst is looking to assess the rela-
tive climate exposure profile to a particular sector across banks, they are unlikely to
find common, audited and detailed climate metrics or assumptions across the board.
Regulators have so far provided little guidance as to what approach(es) should be
preferred and a market standard has yet to emerge. There is however some progress,
notably with disclosures in line with the TCFD recommendations becoming mandatory
in the UK for publicly quoted companies, large private companies and limited liability
partnerships (LLPs) (GOV.UK, 2021).
In June 2021, the FCA issued a consultation on proposals to extend mandatory TCFD
reporting to (i) asset managers, life insurers, FCA-regulated pension providers; and
(ii) issuers of standard listed equity shares, as part of the UK’s ambition to have fully
phased-in TCFD reporting by 2025 (FCA, 2021). This comes as an extension to the
requirement from January 2021 for all premium listed UK companies to comply with
TCFD recommendations and disclose how they are considering the impacts of climate
change, and on a comply or explain basis report against the TCFD framework (FCA,
2021). A phased implementation between January 2022 and January 2023 is sched-
uled, with a final policy statement expected in Q4 2021.
As already mentioned above, in March 2021 the EBA launched a consultation propos-
ing draft Implementing Technical Standards (ITSs) for comparable quantitative disclo-
sures and KPIs on transition and physical risks, and the establishment of a Green Asset
Ratio (EBA, 2021). The consultation closed in June 2021.
The International Financial Reporting Standards Foundation (IFRS) formed a working
group in March 2021 (IFRS, 2021) “to accelerate convergence in global sustainability
reporting standards focused on enterprise value and to undertake technical prepa-
ration for a potential international sustainability reporting standards board under the
governance of the IFRS Foundation”. Its goal is to provide technical recommendations,
based on the TCFD recommendations, as a potential basis for the new board to build
on existing initiatives and develop standards for climate-related reporting and other
sustainability topics.
Recommendation. Early in the process, consider adhering to international stan-

dards and frameworks in climate-related financial disclosures. Adopting and
embedding the available reporting standards at the earliest possible opportunity
as they emerge, will be key to reaping the rewards of this work. Banks should
look to adhere to alignment and disclosure initiatives on a voluntary basis, in
anticipation of their adoption by regulators and before they become mandatory.
Beyond the mandatory requirements for external disclosures, banks should look
to develop their internal dashboards based on their internal climate stress tests,
metrics and targets.
8. Conclusion
and suggested
enhancements
8.1 Conclusion
With the growing recognition of the importance of climate stress testing by the finan-
cial sector, an increasing number of financial institutions are taking part in climate
stress test exercises to identify and assess their resilience to physical and transition
climate risks and to support the firm’s strategy and planning. For financial institutions,
there are four key applications of current climate stress tests:
◾ Mobilising and raising the firm’s awareness of climate-related risks;
◾ Encouraging boards to understand the challenges of climate change and take a
strategic, long-term approach to managing climate risks;
◾ Developing a firm’s capabilities to integrate climate risk into their risk measurement,
assessment and management tools and frameworks; and
◾ Improving climate-related disclosures of financial institutions.
This report identifies obstacles currently faced by institutions and provides recommen-
dations to help overcome these challenges for four specific components of the test—
team organisation and skills, data requirements and collection, selecting scenarios and
models, and meaningful applications of climate stress testing outputs.
Climate stress testing is a much broader exercise than traditional stress testing and
therefore requires institution-wide team engagement, with firms assigning responsibil-
ities to a large number of teams across the organisation. For an effective execution of
a climate stress test, firms should devote appropriate resources (financial and human),
with a specific focus on the activities of data collection and analysis, model develop-
ment and strategic planning. A large body of knowledge and skills are needed in-house
to reach a satisfactory level of proficiency in running the test. Firms should therefore
gear their climate-related training and knowledge development programmes towards a
range of teams across the organisation. Substantial organisational changes should be
made to run the climate stress test, including the development of a climate risk team.
With adequate resources, the climate risk team should have executive sponsorship and
authority to oversee, coordinate and manage the climate stress test. Firms should also
integrate climate stress testing into the firm’s organisational structure and processes
with robust governance and oversight.
Conclusion and suggested enhancements
Climate stress testing requires new kinds of data that were previously not needed for
traditional stress tests. However, when collecting climate-related data, such as emis-
sions data, financial institutions face numerous obstacles. To improve their data collec-
tion capabilities for a climate stress test, firms should develop policies and procedures
to support the collection, processing and use of climate data. Good practices for
collecting data include firms identifying data needs and available sources, implement-
ing industry standards, implementing data validation processes, identifying data gaps
and adapting institutional systems to process these new kinds of data. To overcome
limitations for securing climate data on and from clients, this report recommends firms
to follow best practice steps that they can implement in-house and through collabo-
rations, steps requiring client engagement and steps to build robust data collection
processes. For collecting emissions data, firms should also identify internal sources
through which emissions data for clients can be accessed and develop internal tools
to calculate emissions data. When collecting data from external data providers, it is
recommended that firms develop an internal policy, identify easily accessible open-
source platforms, develop a questionnaire based on the institution’s data needs and
Selecting scenarios and models to use for climate stress testing is a multi-step
process. It is important for firms to adapt reference scenarios geographically and
sector-wise and to derive the impact on key drivers to develop a relevant scenario
narrative. Firms should also implement a process to integrate the latest scenario devel-
opments and emerging methodology standards into the scenario design for climate
stress testing. Financial institutions may also be required to undertake scenario expan-
sion to extrapolate additional scenario variables that may be required for the climate
stress test. The modelling process in a climate stress test consists of modelling
climate variables, measuring the impact of climate risks on macroeconomic variables,
breaking down the macroeconomic impacts per sector or per portfolio and quantifying
the impact on the financial institution. To improve modelling capabilities, a firm should
identify gaps in current in-house models and increase communication with external
modelers and academics on scenario expansion to help improve their understanding
of different models.
Supervisory climate stress tests have a wide range of applications, as mentioned
above. A climate stress test approach should initially be kept simple and focused on
material exposures. However, going forward as the financial sector further expands
climate stress testing, there are six additional uses of a climate stress test:
◾ To better understand vulnerabilities to incorporate into the firm’s risk appetite;
◾ To inform proposed shifts in business strategy;
◾ To inform customer engagement, especially in vulnerable sectors;
◾ To provide direction for the institution’s climate data and infrastructure investments;
◾ To provide value-adding management information and informative external report-
ing and disclosures; and
◾ To develop mitigation strategies and push further analysis.
8.2 Next steps for climate stress testing
Understanding the current landscape of climate stress testing, it is important to look
at the future of climate stress testing. This section will discuss the current limitations
of the test and suggest ways for supervisors to continue developing the exercise in
the future. Below we highlight some of the key issues that have been identified in this
report that need to be addressed by the financial sector to allow for an effective execu-
tion of a climate stress test.
Lack of skills and internal expertise

Executing a climate stress test requires teams across the organisation to be proficient
in a range of climate-related skills. However, many teams currently lack the skills and
knowledge needed to understand and assess the financial impacts of climate risks.
Often the majority of climate-related training at firms is geared towards client-fac-
ing employees. However, as climate stress tests require the collaboration of multiple
teams from across the organisation, it is important for institutions to build internal
expertise on climate risks and climate stress testing much more widely.
Lack of data availability

Currently firms find it difficult to obtain the necessary client-related data since a large
proportion of clients do not provide comprehensive and complete climate data disclo-
sures. Furthermore, the financial sector has not yet reached a consensus on the types
of physical and transition risk data that should be used for a climate stress test.
High costs and resource intensity

Climate stress testing is a highly resource-intensive process with firms needing to
devote a sufficient amount of resources and capital for designing, implementing and
executing the exercise. This as a result can be costly. For example, counterparty anal-
ysis is resource-intensive and can be expensive to implement. Furthermore, firms may
also rely on third-parties to support them in various activities, such as data collection,
modelling and auditing, which can also be costly.
Climate scenario selection

To increase relevance of scenarios for climate stress tests, scenarios need to be
adapted with features relevant to the firm’s business model. As climate scenarios
were not initially designed for use by the financial sector, they may not sufficiently link
climate drivers to economic impacts for analysis by financial institutions.
Limited model design and their uncertainty

Models used for climate stress testing were not initially designed for the purpose of
translating climate risks into financial impacts for financial institutions. As a result,
when executing a climate stress test, firms face many challenges, including the uncer-
tainty of modelling outputs. These complex models are run on long time horizons with
multiple assumptions which lead to uncertainties.
Based on this, the box below lists guidance for regulators as climate stress tests are
further developed.
Seven recommendations for regulators to assist firms in their

climate stress testing journey
1. Help build institutional expertise on climate change and climate risks by
providing workshops, trainings and certified courses for firms.
2. Require mandatory disclosures by clients on necessary data needed for
climate stress testing, such as emissions data.
3. Implement policies on disclosing required data, like emissions data, in stan-
dardised formats to allow for comparability. Standardised global taxonomies
will allow firms to collect comparable data across sectors and regions.
4. Provide free open-source data to firms for a climate stress test.
5. Engage with financial institutions to modify current systems used to run
models to make them suitable for running climate stress tests.
6. Engage with financial institutions to design scenarios for supervisory
climate stress tests that are relevant to the firm’s scope and business model,
in terms of risks and exposures assessed, time horizon, regions and sectors.
7. Initiate communication between financial institutions and climate scientists,
academic and research institutions, modelers and tool providers to support
firms in building their capacity for climate stress tests.
Though still in its early stages, climate stress tests are continuing to grow as a tool
used by the finance sector for risk assessment. With the growing threat of climate
change and increasing pressure by supervisory authorities, an increasing number
of financial institutions are beginning to conduct climate stress tests to assess their
resilience to physical and transition risks. As climate stress tests are more broadly
implemented, central banks, prudential regulators and various initiatives around the
world are continuing their work to address the limitations of climate stress tests and
expand the scope of the exercise. UNEP FI will continue to work with our partners and
members across the financial sector to further develop sophisticated approaches to
climate stress testing and scenario analysis as part of our TCFD programme, helping
financial institutions build their capabilities to assess climate risks and develop mean-
ingful climate risk disclosures.
9. Appendices
Appendix 1
Key questions and topics to address when collecting physical risk data from clients:
These questions and topics can be addressed in the form of a questionnaire.
◾ What kinds of climate hazards is the client typically exposed to? Hazards can
include flooding, drought, cyclones etc.
◾ Does the client have adaptation and resilience plans relevant to these climate
hazards? If so, what data is available on these plans?
◾ Does the client have data available on previous exposures to extreme weather
events and their impacts?
◾ Description of the physical assets owned by the client and information on their
geographical location, including address, postal code, coordinates and availability of
geographical maps.
◾ What methodologies have been used by clients to gather the data and what are the
underlying assumptions and limitations of these methodologies?
Appendix 2
Key questions and topics to address when collecting transition risk data from clients:
These questions and topics can be addressed in a questionnaire.
◾ Does the client have transition plans, if so, what data is available on the transition
plans and pathways to reduce emissions? If the client does not have transition plans
in place, how and when do they intend to develop them?
◾ Does the client have data available on plans for decarbonisation targets or pledges?
◾ Is the client part of a low-carbon transition initiative and if so, which one? Does the
client plan to join any initiatives in the future? Examples of some of these initiatives
are provided in Figure 11.
◾ What data does the client have available to determine its carbon sensitivity? Data
can include its current GHG emissions, the client’s productive capacity in relation to
its emission production and a client’s ability to decrease its emissions.
Appendices
◾ Has the client identified its exposure to carbon prices? If so, what data and/or
assumptions have been used to quantify it?
◾ What kinds of data does the client have available on energy efficiency and mix?
◾ What methodologies have been used by the client to gather the data and what are
the underlying assumptions and limitations of these methodologies?
Appendix 3
Additional questions for collecting emissions data include:
◾ What kinds of data are available on a client’s emission-releasing activity?
◾ Which of the six GHGs from the Kyoto Protocol are emitted by the client? What kinds
of data are available on these GHGs?
◾ What expenditure data is available regarding client purchases of goods or services?
This data will be useful in estimating the emissions of a client.
◾ Identification of sources used by the client to obtain data and their limitations, since
the PCAF (PCAF, 2020) recommends that emissions data be reported using a stan-
dardised framework, for example the CDP framework or the firm’s official disclo-
sures and environmental reports.
◾ What methodologies have been used by the client to gather the data and what are
the underlying assumptions and limitations of these methodologies? Clients and
data providers can have their own methodologies for estimating GHG emissions
(PCAF, 2020).
Appendix 4
Key questions to address when gathering physical risk data from external providers:
◾ Does the external data provider provide data on historical extreme weather events?
◾ Does the external data provider provide data on future projections for extreme
weather events?
◾ What is the geographical coverage and resolution of the data available and does the
data provider cover regions of interest for the financial institution?
◾ What geographical data on the location is available, including the type of land, eleva-
tion, composition of soil and land cover?
◾ Does the data provider cover sectors of interest for the financial institution?
◾ Which sources and methodologies are used by the data provider to obtain the data?
What are the assumptions and limitations of these sources and methodologies?
Appendices
Appendix 5
Key questions to address when gathering transition risk data from external providers:
◾ Does the external data provider provide data on carbon pricing by jurisdiction?
◾ Does the external data provider provide granular data on Scope 1, 2 and 3 GHG
emissions?
◾ Does the external data provider provide data on pledges, targets and commitments?
◾ What is the geographical coverage of the data available and does the data provider
cover regions of interest for the financial institution?
◾ What is the sectoral coverage of the data available and does the data provider cover
sectors of interest for the financial institution?
◾ Which sources and methodologies are used by the data provider to obtain the data?
◾ What are the assumptions and limitations of these sources and methodologies?
Appendices
10. Glossary
Climate-related risk
The potential negative impacts of climate change on a financial institution.
Credit Risk
Possibility of loss resulting from borrower’s inability to repay loans.
Stranded Assets
The premature write-off of assets incompatible with a sustainable economy.
GHG Protocol
The GHG Protocol establishes comprehensive global standardised frameworks to
measure and manage greenhouse gas emissions from private and public sector opera-
tions, value chains and mitigation actions.
Sustainable Finance
The provision of finance to investments taking into account environmental, social and
governance considerations.
Bottom-up
A bottom-up approach to climate stress testing is when a firm uses its own framework
as part of a system-wide or supervisory exercise.
Climate-related disclosures
Disclosures that promote more informed investment, credit and insurance underwrit-
ing decisions, vis-à-vis climate risk, and in turn enable stakeholders to better under-
stand the concentrations of carbon-related assets in the financial sector and the
financial system’s exposures to climate-related risks.
Climate stress test

A climate stress test is a forward-looking exercise designed to measure a financial
institution’s exposure to climate risks, using scenario analysis including severe climate
risks, to assess the potential impact of climate change on the institution’s business
model.
Counterparty
A counterparty is the other party participating in a transaction, which could be a legal
entity, unincorporated entity or collection of entities to which an exposure of financial
risk may exist.
Glossary
Exposure at Default
Exposure at default is the total value a bank is exposed to when a loan defaults. Expo-
sure at default is calculated taking account of the underlying asset, forward valuation,
facility type and commitment details.
Geographic Information Systems

A GIS is a computer system that enables the capturing, analysis and display of spatial
and geographically referenced data.
Liquidity risk
Liquidity risk is a financial risk that for a certain period of time a given financial asset,
security or commodity cannot be traded quickly enough in the market without impact-
ing the market price. Climate risks may impact a bank’s and its clients’ abilities to raise
funds and liquidate assets.
Litigation risk
Litigation risk is the risk an individual or company will face legal action.
Loss Given Default

Loss given default is the share of an asset that is lost if a borrower defaults.
Market risk
Market risk is the possibility that an individual or other entity will experience losses due
to market volatility and change in asset prices, such as interest rate risk, commodity
price risk, corporate bond spreads, and equity price risk.
Net zero
Net zero is shorthand for ‘net-zero CO2 emissions by 2050’ and is derived from the
science-based definition of net zero which means that all accumulated emissions over
the next 30 years remain within the emissions budget required and set out by the IPCC
to align with a 1.5°C pathway (stabilising planetary warming at no more than 1.5°C
above pre-industrial levels by 2100–50 years after net zero is achieved).
NiGEM
NiGEM is a leading global macroeconomic model maintained by NIESR used to create
country economic forecasts, scenarios and simulations, allowing the evaluation of the
impact of specific economic shocks and alternative scenarios.
Operational risk
Risk of loss resulting from ineffective internal processes that can disrupt the flow of
business operations, caused by uncontrollable events, gaps in business flows, damage
to physical assets, and/or business disruptions.
Glossary
Physical risk
Physical risks are risks associated with the impact of changes in weather and
climate, on the economy. Physical risks can be acute in nature, like floods, heatwaves
and wildfires, or chronic in nature, such as sea level rise, temperature changes and
precipitation changes.
Probability of Default
Probability of default is the risk/likelihood that the borrower will be unable or unwilling
to repay its debt in full or on time.
Risk Appetite
Risk appetite is the amount of risk that an institution is willing to accept in order to
achieve its objectives.
Risk-weighted asset
Risk-weighted assets are used to determine the minimum amount of capital needed to
be held by banks in order to maintain solvency.
Supervisory authority
An independent public authority responsible for monitoring the application of regula-
tion and key supervisory frameworks.
Transition risk
Transition risks are related to changes in legislation, policies, technology and the
market, during the transition towards a low-carbon economy.
Traditional stress test

Traditional stress tests are an integral part of a financial institution’s risk management
toolkit with regulators conducting large-scale supervisory stress tests. A traditional
stress test is used to measure the capital resiliency of a financial institution to severe,
hypothetical scenarios. The results of a stress test are then used by central banks
and regulators to understand risks, adjust capital requirements and develop policy for
financial resilience.
Top-down
A top-down approach is when a supervisory authority performs the test themselves,
using their own framework, which includes a homogenised methodology, assumptions,
scenarios and models.
Glossary
11. Bibliography
ACPR (2020). Scenarios and main assumptions of the ACPR pilot climate exercise.
Banque de France. Available at: acpr.banque-france.fr/sites/default/files/medias/docu-
ments/20200717_main_assumptions_and_scenarios_of_the_acpr_climate_pilot_exer-
cise.pdf. [Accessed: 05.10.2021]
ACPR (2021). A first assessment of financial risks stemming from climate change: The
main results of the 2020 climate pilot exercise. Banque de France. Available at: acpr.
banque-france.fr/sites/default/files/medias/documents/20210602_as_exercice_pilote_
english.pdf. [Accessed: 05.10.2021]
BIS (2021). Climate-related financial risks—measurement methodologies. Bank for Inter-
national Settlement. Available at: bis.org/bcbs/publ/d518.pdf. [Accessed: 05.10.2021]
Bank of Canada (2020). Bank of Canada and OSFI launch pilot project on climate risk
scenarios. Available at: bankofcanada.ca/2020/11/bank-canada-osfi-launch-pilot-proj-
ect-climate-risk-scenarios/. [Accessed: 11.10.2021]
Bank of England (2016). Stress testing of banks: an introduction. Available at: bankofen-
gland.co.uk/-/media/boe/files/quarterly-bulletin/2016/stress-testing-of-banks-an-intro-
duction.pdf. [Accessed: 11.10.2021]
Bank of England (2019). Enhancing banks’ and insurers’ approaches to managing the
financial risks from climate change. Available at: bankofengland.co.uk/prudential-regu-
lation/publication/2019/enhancing-banks-and-insurers-approaches-to-managing-the-
financial-risks-from-climate-change-ss. [Accessed: 05.10.2021]
Bank of England (2019). Supervisory Statement | SS3/19 Enhancing banks’ and insurers’
approaches to managing the financial risks from climate change. Available at: bankofen-
gland.co.uk/prudential-regulation/publication/2019/enhancing-banks-and-insurers-
approaches-to-managing-the-financial-risks-from-climate-change-ss. [Accessed:
05.10.2021]
Bank of England (2019). The 2021 biennial exploratory scenario on the financial risk
from climate change. Available at: Discussion Paper: The 2021 biennial exploratory
scenario on the financial risks from climate change (bankofengland.co.uk). [Accessed:
05.10.2021]
Bank of England (2020). Update on the Bank’s approach to the Climate Biennial Explor-
atory Scenario in selected areas. Available at: bankofengland.co.uk/-/media/boe/files/
stress-testing/2020/update-on-the-banks-approach-to-the-climate-biennial-explorato-
ry-scenario. [Accessed: 05.10.2021]
Glossary
Bank of England (2021). Guidance for participants of the 2021 Biennial Explor-
atory Scenario: Financial risks from climate change. Available at: bankofengland.
co.uk/-/media/boe/files/stress-testing/2021/the-2021-biennial-exploratory-sce-
nario-on-the-financial-risks-from-climate-change.pdf?la=en&hash=2E5CAE-
CE75E701315B51B09303F99FCF8D21C8E2. [Accessed: 05.10.2021]
Bank of England (2021). Key elements of the 2021 biennial exploratory scenario: Finan-
cial risks from climate change. Available at: bankofengland.co.uk/stress-testing/2021/
key-elements-2021-biennial-exploratory-scenario-financial-risks-climate-change.
[Accessed: 05.10.2021]
Banque de France (2020). Climate-Related Scenarios for Financial Stability Assessment:
an Application to France. Banque de France. Available at: publications.banque-france.fr/
sites/default/files/medias/documents/wp774.pdf. [Accessed: 05.10.2021]
BIS (2009). Principles for sound stress testing practices and supervision. Available at:
bis.org/publ/bcbs155.pdf. [Accessed: 05.10.2021]
BIS (2020). Calculation of RWA for credit risk. Bank for International Settlement. Avail-
able at: bis.org/basel_framework/chapter/CRE/53.htm. [Accessed: 05.10.2021]
BIS (2021). Stress-testing banks for climate change – a comparison of practices. Bank
for International Settlement. Available at: bis.org/fsi/publ/insights34.htm. [Accessed:
05.10.2021]
BPI (2020). Challenges in Stress Testing and Climate Change. Available at: bpi.com/
challenges-in-stress-testing-and-climate-change/. [Accessed: 05.10.2021]
Carbonbrief (2019). Explainer: The high-emissions RCP8.5 global warming scenario.
Available at: carbonbrief.org/explainer-the-high-emissions-rcp8-5-global-warming-sce-
nario. [Accessed: 05.10.2021]
CDC (2021). Climate Effects on Health. Available at: cdc.gov/climateandhealth/effects/
default.htm. [Accessed: 11.10.2021]
CICERO (2017). Shades of Climate Risk: Categorizing climate risk for investors.
Available at: pub.cicero.oslo.no/cicero-xmlui/bitstream/handle/11250/2430660/
Shades%20of%20Climate%20Risk%20nyformatting%20(Sistetil%20web13%2002%20
2016).pdf?sequence=1&isAllowed=y. [Accessed: 12.10.2021]
CISL (2021) The essential guide to bank-client engagement. Cambridge University.
Available at: cisl.cam.ac.uk/resources/sustainable-finance-publications/lets-discuss-
climate-the-essential-guide-to-bank-client-engagement. [Accessed: 05.10.2021]
Climate Risk Review (2020). One direction: which way to effective climate stress tests?
Available at: climateriskreview.com/p/one-direction-which-way-to-effective. [Accessed:
05.10.2021]
Danmarks Nationalbank (2020). A gradual green transition supports financial stabil-
ity. Available at: nationalbanken.dk/en/publications/Documents/2020/11/ANALY-
SIS_No.21_A%20gradual%20green%20transition%20supports%20financial%20stability.
pdf#search=climate. [Accessed: 05.10.2021]
Deloitte (2020). Climate change stress testing. Available at: www2.deloitte.com/cn/en/
pages/risk/articles/climate-change-stress-testing.html. [Accessed: 11.10.2021]
Glossary
De Nederlandsche Bank (2018). An energy transition risk stress test for the financial
system of the Netherlands. Available at: dnb.nl/media/pdnpdalc/201810_nr-_7_-2018-_
an_energy_transition_risk_stress_test_for_the_financial_system_of_the_netherlands.
pdf. [Accessed: 05.10.2021]
EBA (2018). Final report: Guidelines on institutions’ stress testing. EBA. Available at: eba.
europa.eu/sites/default/documents/files/documents/10180/2282644/2b604bc8-
fd08-4b17-ac4a-cdd5e662b802/Guidelines%20on%20institutions%20stress%20
testing%20(EBA-GL-2018-04).pdf?retry=1. [Accessed: 05.10.2021]
EBA (2018). Guidelines on stress testing. Available at: eba.europa.eu/regula-
tion-and-policy/supervisory-review-and-evaluation-srep-and-pillar-2/guidelines-on-
stress-testing2. [Accessed: 05.10.2021]
EBA (2021). EBA launches public consultation on draft technical standards on Pillar 3
disclosures of ESG risks. Available at: eba.europa.eu/eba-launches-public-consulta-
tion-draft-technical-standards-pillar-3-disclosures-esg-risks. [Accessed: 05.10.2021]
EBA (2021). EBA published its report on management and supervision of ESG risks
for credit institutions and investments firms. Available at: eba.europa.eu/eba-publish-
es-its-report-management-and-supervision-esg-risks-credit-institutions-and-invest-
ment. [Accessed: 05.10.2021]
ECB (2020). Guide on climate-related and environmental risks: Supervisory expectations
relating to risk management and disclosure. European Central Bank Banking Supervi-
sion. Available at: bankingsupervision.europa.eu/ecb/pub/pdf/ssm.202011finalguide-
onclimate-relatedandenvironmentalrisks~58213f6564.en.pdf. [Accessed: 05.10.2021]
ECB (2021). Climate-related risk and financial stability. European Central Bank. Available
at: ecb.europa.eu/pub/pdf/other/ecb.climateriskfinancialstability202107~87822fae81.
en.pdf. [Accessed: 05.10.2021]
ECB (2021). Patchy data is a good start: from Kuznets and Clark to supervisors
and climate. European Central Bank. Available at: bankingsupervision.europa.eu/
press/speeches/date/2021/html/ssm.sp210616~a39a86eeb2.en.html. [Accessed:
05.10.2021]
ECB (2021). Shining a light on climate risks: the ECB’s economy-wide climate stress
test. European Central Bank. Available at: ecb.europa.eu/press/blog/date/2021/html/
ecb.blog210318~3bbc68ffc5.en.html. [Accessed: 05.10.2021]
ECB (2021). Occasional Paper Series. ECB economy-wide climate stress test. Methodol-
ogy and results. European Central Bank. Available at: ecb.europa.eu/pub/pdf/scpops/
ecb.op281~05a7735b1c.en.pdf. [Accessed: 11.10.2021]
FCA (2020). Guide 2020 Scenario Analysis Chapter. Available at: fca.org.uk/publica-
tion/corporate/climate-financial-risk-forum-guide-2020-scenario-analysis-chapter.pdf.
[Accessed: 05.10.2021]
FCA (2020). Risk Management Chapter. Available at: fca.org.uk/publication/corporate/
climate-financial-risk-forum-guide-2020-risk-management-chapter.pdf. [Accessed:
05.10.2021]
Glossary
FCA (2021). Climate-related reporting requirements. Available at: fca.org.uk/firms/
climate-change-sustainable-finance/reporting-requirements. [Accessed: 05.10.2021]
FCA (2021). FCA consults on further climate-related disclosure rules. Available at: fca.
org.uk/news/press-releases/fca-consults-further-climate-related-disclosure-rules.
[Accessed: 05.10.2021]
Finbarr Bermingham (2021). China’s central bank chief Yi Gang says climate-related
stress test results will be released. South China Morning Post. Available at: scmp.
com/knowledge/china-macro-economy?module=section_masthead&pgtype=article.
[Accessed: 05.10.2021]
FSB (2017). Technical Supplement. The use of scenario analysis in disclosure of
climate-related risks and opportunities. Available at: Technical-Supplement-1.pdf (fsb.
org). [Accessed: 05.10.2021]
GOV.UK (2021). Mandatory climate-related financial disclousres by publicly quoted
companies, large private companies and LLP’s. Available at: gov.uk/government/consul-
tations/mandatory-climate-related-financial-disclosures-by-publicly-quoted-compa-
nies-large-private-companies-and-llps. [Accessed: 05.10.2021]
Greenhouse Gas Protocol (2021). Calculation tool. Available at: ghgprotocol.org/
ghg-emissions-calculation-tool. [Accessed: 05.10.2021]
HKMA (2020). Climate risk stress test. Available at: hkma.gov.hk/media/eng/
doc/key-information/guidelines-and-circular/2020/20201204e1.pdf. [Accessed:
05.10.2021]
HKMA (2020). Range of practices for management of climate risks. Available at: hkma.
gov.hk/media/eng/doc/key-information/guidelines-and-circular/2020/20200707e1a1.
pdf. [Accessed: 05.10.2021]
I4CE (2021). The integration of transition risk drivers at a sectoral level: new perspec-
tives for climate stress tests and other climate risk management tools. Institute for
Climate Economics. Available at: i4ce.org/wp-core/wp-content/uploads/2021/07/
Rapport-IntegrationFacteurRisque-VA.pdf. [Accessed: 11.10.2021]
IFRS (2021). IFRS Foundation Trustees announce working group to accelerate conver-
gence in global sustainability reporting standards focused on enterprise value. Avail-
able at: ifrs.org/news-and-events/news/2021/03/trustees-announce-working-group/.
[Accessed: 05.10.2021]
IIF (2021). Navigating Climate Headwinds: Reference Approaches for scenario-based
Climate Risk Measurement by banks and supervisors. Available at: iif.com/Publi-
cations/ID/4506/Navigating-Climate-Headwinds-Reference-Approaches-for-Sce-
nario-based-Climate-Risk-Measurement-by-Banks-and-Supervisors. [Accessed:
05.10.2021]
International Energy Agency (2020). World Energy Model. Available at: World Energy
Model—Analysis—IEA. [Accessed: 05.10.2021]
International Energy Agency (2020). World Energy Model Documentation. Available at:
iea.blob.core.windows.net/assets/bc4936dc-73f1-47c3-8064-0784ae6f85a3/WEM_
Documentation_WEO2020.pdf. [Accessed: 05.10.2021]
Glossary
International Energy Agency (2021). Net zero by 2050. Available at: Net Zero by 2050—
Analysis—IEA. [Accessed: 05.10.2021]
International Energy Agency (2021). Sustainable Development Scenario. Available
at: iea.org/reports/world-energy-model/sustainable-development-scenario#:~:-
text=The%20IEA%E2%80%99s%20Sustainable%20Development%20Scenario%20
%28SDS%29%20outlines%20a,deliver%20on%20the%20three%20main%20energy-re-
lated%20SDGs%20simultaneously. [Accessed: 12.10.2021]
Investopia (2020). Exposure at Deafault. Available at: investopedia.com/terms/e/expo-
sure_at_default.asp. [Accessed: 05.10.2021]
IPCC (2021). Future Climate Changes, Risks and Impacts. Available at: ar5-syr.ipcc.ch/
topic_futurechanges.php. [Accessed: 05.10.2021]
IPCC (2021). Climate Change 2021: The Physical Science Basis. Available at: ipcc.ch/
report/sixth-assessment-report-working-group-i/. [Accessed: 12.10.2021]
Laura Singleton (2020). Business School establishes new Singapore Green Finance
Centre. Imperial College London. Available at: imperial.ac.uk/news/206473/busi-
ness-school-establishes-singapore-green-finance/.[Accessed: 05.10.2021]
MAS (2021). “Being the Change We Want to See: A Sustainable Future”. Speech by Mr
Ravi Menon, Managing Director, Monetary Authority of Singapore, at Launch of Inaugural
MAS Sustainability Report via Video Conference on 9 June 2021. Available at: mas.gov.
sg/news/speeches/2021/a-sustainable-future. [Accessed: 05.10.2021]
NGFS (2020). Guide to climate scenario analysis for central banks and supervisors.
Available at: ngfs.net/sites/default/files/medias/documents/ngfs_guide_scenario_
analysis_final.pdf. [Accessed: 05.10.2021]
NGFS (2021). NGFS climate scenarios for central banks and supervisors. Available
at: ngfs.net/sites/default/files/media/2021/08/27/ngfs_climate_scenarios_phase2_
june2021.pdf. [Accessed: 05.10.2021]
NGFS (2021). Progress report on bridging data gaps. Available at: ngfs.net/sites/
default/files/medias/documents/progress_report_on_bridging_data_gaps.pdf.
[Accessed: 05.10.2021]
NGFS (2021). Scenarios in Action: a progress report on global supervisory and central
bank climate scenario exercises. Available at: ngfs.net/en/scenarios-action-prog-
ress-report-global-supervisory-and-central-bank-climate-scenario-exercises. [Accessed:
20.10.2021]
OSFI (2009). Stress Testing. Government of Canada. Available at: osfi-bsif.gc.ca/Eng/
fi-if/rg-ro/gdn-ort/gl-ld/Pages/e18.aspx. [Accessed: 05.10.2021]
Open Risk Manuel (2016). Bottom-Up versus Top-Down Stress Test. Available at:
openriskmanual.org/wiki/Bottom-Up_versus_Top-Down_Stress_Test. [Accessed:
11.10.2021]
PCAF (2020). The Global GHG Accounting & Reporting Standard. Available at: carbonac-
countingfinancials.com/files/downloads/PCAF-Global-GHG-Standard.pdf. [Accessed:
05.10.2021]
Glossary
Pierpaolo Grippa, Jochen Schmittmann, and Felix Suntheim (2019). Climate Change
and Financial Risk. International Monetary Fund. Available at: imf.org/external/pubs/
ft/fandd/2019/12/pdf/climate-change-central-banks-and-financial-risk-grippa.pdf.
[Accessed: 11.10.2021]
Pierpaolo Grippa, Samuel Mann (2021). Climate-Related Stress Testing: Transitions
Risks in Norway. International Monetary Fund. Available at: imf.org/en/Publications/
WP/Issues/2020/11/08/Climate-Related-Stress-Testing-Transition-Risks-in-Nor-
way-49835. [Accessed: 11.10.2021]
RCP Database (2021). Version 2.0. Available at: tntcat.iiasa.ac.at/RcpDb/dsd?Ac-
tion=htmlpage&page=welcome. [Accessed: 05.10.2021]
SSRN (2020). Climate-related scenarios for financial stability assessment: An applica-
tion to France. Available at: papers.ssrn.com/sol3/papers.cfm?abstract_id=3653131.
[Accessed: 05.10.2021]
TCFD (2017). Recommendations of the Task Force on Climate-related Financial Disclo-
sures. Available at: assets.bbhub.io/company/sites/60/2020/10/FINAL-2017-TCFD-
Report-11052018.pdf. [Accessed: 05.10.2021]
TCFD (2017). The use of Scenario analysis in disclosure of Climate-related risks and
opportunities. Available at: FINAL-TCFD-Technical-Supplement-062917.pdf (bbhub.io).
[Accessed: 05.10.2021]
TCFD (2020). Guidance on Risk Management Integration and Disclosure. TCFD. Avail-
able at: assets.bbhub.io/company/sites/60/2020/09/2020-TCFD_Guidance-Risk-Man-
agement-Integration-and-Disclosure.pdf. [Accessed: 05.10.2021]
TCFD (2021). Forward-looking financial metrics consultation. Available at: assets.bbhub.
io/company/sites/60/2021/03/Summary-of-Forward-Looking-Financial-Metrics-Con-
sultation.pdf. [Accessed: 05.10.2021]
Thomson et. Al, (2011). RCP4.5: A Pathway for Stabilization of Radiative Forcing
by 2100. Atmospheric System Research. Available at: ASR—Atmospheric System
Research (energy.gov). [Accessed: 05.10.2021]
UNEP FI (2021). Pathways to Paris. Available at: unepfi.org/wordpress/wp-content/
uploads/2021/02/UNEP-FI-Pathways-to-Paris.pdf. [Accessed: 05.10.2021]
Van Vuuren et. al., (2011). The representative concentration pathways: an overview.
Springer Link. Available at: link.springer.com/article/10.1007/s10584-011-0148-z.
[Accessed: 05.10.2021]
World Resources Institute & World Business Council for Sustainable Development
(2004). GHG Protocol Corporate Accounting and Reporting Standard. Available at:
ghgprotocol.org/sites/default/files/standards/ghg-protocol-revised.pdf. [Accessed:
05.10.2021]
Glossary
United Nations Environment Programme Finance Initia-
tive (UNEP FI) is a partnership between UNEP and the
global financial sector to mobilise private sector finance
for sustainable development. UNEP FI works with more
than 400 members—banks, insurers, and investors—and
over 100 supporting institutions– to help create a financial
sector that serves people and planet while delivering posi-
tive impacts. We aim to inspire, inform and enable finan-
cial institutions to improve people’s quality of life without
compromising that of future generations. By leveraging the
UN’s role, UNEP FI accelerates sustainable finance.
unepfi.org
unepfi.org
[email protected]
/UNEPFinanceInitiative
United Nations Environment Finance Initiative
@UNEP_FI

Good Practice Guide To Climate Stress Testing

Uploaded by

Copyright:

Available Formats

Good Practice Guide To Climate Stress Testing

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Good Practice Guide To Climate Stress Testing

Uploaded by

Copyright:

Available Formats

UNEP FI’s

Table 1: Comparison of key features of traditional and climate stress testing....................................... 27

Exhibit 1: Monetary Authority of Singapore’s Action Plan to build knowledge and

ACPR Autorité de contrôle prudentiel et de résolution (France)

1.2.1 Team organisation and skills

Climate stress testing is a full institutional endeavor. It demands engagement from

Scenario design Financial risk assessment Outcome

Counterparty credit risk

Internal audit and Change Standard modelling structure

Physical risk variables Transition risk variables

Financial Models Loss Given Default (LGD) Probability of Default (PD)

Probability weighted Potential impact of climate

Figure 2: Modelling Approach for Climate Stress Testing.

1.2.4 Outputs and applications

2.1 Importance of addressing climate risks

2.2 History of stress testing

2.3 Overview of the UNEP FI TCFD programme

3.1 Comparison of stress tests

3.1.1 Purpose of traditional and climate stress tests

3.1.2 Comparing key features of traditional and climate

Characteristic Traditional Stress Test Climate Stress Test

Table 2: Overview of top-down and bottom-up stress testing.

Types of Stress Tests

3.3 Climate stress testing regulatory landscape

3.3.1 List of recent announcements and initiatives

Expected end Balance

4.1.1 Institution-wide team engagement for climate

4.1.2 Teams and their respective roles

Scenario design Financial risk assessment Outcome

Counterparty credit risk

Internal audit and Change Standard modelling structure

Team Team Definition Responsibilities

Recommendation: Firms should devote the appropriate resources (financial

4.2 Knowledge and skills

Country Risk Team

Market Risk Team

Board and Execu-

Country Risk Team

Market Risk Team

Board and Execu-

Knowledge and understanding of various models, such as IAMs, including their

Understanding of the finan-

* Understanding the process and controls around the topics highlighted

Set of recommendations to enhance the development of knowledge

Climate-related training and knowledge development programmes are often geared

4.2.2 Building skills through external support

Recommendation: Firms should leverage external support to develop in-house

4.3 Institutional organisation and governance

4.3.1 Changes to team organisation for climate stress testing

Develop a climate risk team with adequate resources, executive

Collate the results and analyse them in

Figure 5: Recommended roles for the climate risk team.

Establish a climate research and intelligence function

Manage climate-related data centrally

Six recommendations for financial institutions to implement changes

Embedding climate stress testing into existing stress

Robust governance and oversight arrangements

Figure 6: TCFD’s recommendations for governance.

Adequate resourcing for all climate stress testing functions