FINANCE FINANCE EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT Assessing Financial Risks from Physical Climate Shocks: A Framework for Scenario Generation Authors: Nicola Ranger, Olivier Mahul, and Irene Monasterolo © 2022 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Cover design and layout: Diego Catto / www.diegocatto.com >>> Acknowledgments This report was prepared by Dr. Nicola Ranger (Oxford Sustainable Fi- nance Group, Smith School of Enterprise and the Environment, University of Oxford and the UK Centre for Greening Finance and Investment), Dr. Olivier Mahul (Practice Manager, Crisis and Disaster Risk Finance Unit, The World Bank), and Prof. Irene Monasterolo (EDHEC business school and EDHEC-Risk). We wish to thank colleagues for their input and reviews of this paper, including Cathy Ansell, Antoine Bavandi, and Lit Ping Low from the World Bank’s Crisis and Disaster Risk Finance Team. We would like to thank Jean Pesme, Erik Feyen, Florent McIsaac, Moritz Baer, and Martin Me- lecky for their valuable review comments and insights. Any remaining er- rors are those of the authors. We acknowledge Monica Billio, Stefano Bat- tiston, Louison Cahen Fourot, Nepomuk Dunz, Arthur Hrast Essenfelder, Andrea Mazzocchetti, and Malcolm Mistry from Ca’ Foscari University of Venice for their contributions to the wider World Bank research project that generated figures 4.2, 4.3, and 4.4, supported by the Global Risk Fi- nancing Facility. Nicola Ranger acknowledges financial support from the UK Natural Environment Research Council (NERC Grant NE/V017756/1). Olivier Mahul acknowledges financial support from the Disaster Protec- tion Program and the Global Risk Financing Facility. Irene Monasterolo acknowledges the financial support of the 11th Klimafonds+ call [Green- Fin, grant number KR18ACOK14634], of the European Union’s Horizon 2020 research and innovation programme [CASCADES, grant number 821010], and of the European Investment Bank Institute [EIBURS project ESG-Credit.eu]. >>> Abstract Climate change has become a main concern of ministries of finance, cen- tral banks, and financial regulators. In response, a suite of scenarios and tools have been developed to assess the financial risks from physical cli- mate shocks (for example, hurricanes, droughts, wildfires, flooding). How- ever, those scenarios do not fully capture such shocks, which could lead financial institutions to underestimate the potential scale of climate risks and underprice investments in resilience. This is particularly important for emerging markets and developing economies where exposure to physi- cal climate risks is already high and is expected to further increase with climate change. The paper identifies five areas, or risk drivers, that make a material contribution to physical climate risks to the financial sector and that are not consistently included in current scenarios and tools: (1) ex- treme weather events, (2) uncertainties in climate models, (3) compound scenarios, (4) indirect economic impacts of shocks, and (5) feedback be- tween the real economy and the financial sector. We derive a framework for generating scenarios to assess acute physical climate-related financial risks, which is inspired by the “Realistic Disaster Scenarios” that are used in risk management and supervision in the insurance sector. The frame- work is illustrated through an application of the EIRIN macroeconomic model. This framework aims to complement recent work by the Network of Central Banks and Supervisors for Greening the Financial System (NGFS) and the Financial Stability Board (FSB) to inform ministries of finance, central banks, financial regulators, and financial institutions on climate financial risk assessments, both for micro- and macroprudential risk management, and to incorporate climate risks into wider financial de- cision making and disclosures. Keywords: physical climate risk; climate-related financial risk scenarios; risk drivers; macrofinancial feedbacks, macroprudential supervision; risk management; low-income countries. >>> Acronyms AAL Average Annual Loss BAU Business as usual BIS Bank for International Settlements BoE Bank of England CAR Capital Adequacy Ratio CBES Climate Biennial Exploratory Scenario CGE Computable general equilibrium CMIP Climate Model Intercomparison Project DRF Disaster risk finance DSGE Dynamic stochastic general equilibrium ECB European Central Bank EU European Union FSAP Financial Sector Assessment Program FSB Financial Stability Board GCM Global Climate Models GDP Gross domestic product GVA Gross Value Added IAM Integrated Assessment Models IMF International Monetary Fund IPCC Intergovernmental Panel on Climate Change ISIMIP Inter-Sectoral Impact Model Intercomparison Project LAC Latin America and the Caribbean MFI Microfinance institution NGFS Network of Central Banks and Supervisors for Greening the Financial System NPL Nonperforming Loan NRB Nepal Rastra Bank RCP Representative Concentration Pathways SACCO Savings and Credit Cooperatives SFC Stock-flow consistent SIDS Small island developing states >>> Contents Acknowledgements 3 Abstract 4 Abbreviations and Acronyms 5 1. Introduction 7 2. Review of Current Scenarios for Physical Climate Risks 10 3. Empirical Evidence on the Impacts of Physical Climate Shocks to Banks 18 4. Revisiting the Development of Physical Climate Financial Risk Scenarios 22 4.1. Scenario Generation for Extreme Weather Events Under Current and Future Climate Change 23 4.2. Fully Capturing the Uncertainty in Current Climate and Impact Projections 25 4.3. Representing the Indirect Impacts of Physical Climate-Related Shocks 26 4.4. Impacts on the Financial Sector and Economy-Financial Sector Feedbacks 27 4.5. Compounding Risks 33 5. Next Steps Toward Implementation 35 6. Conclusions 38 References 40 Figure Figure 1.1: Approach to Scenario Generation and Analysis Proposed by the Climate Financial 9 Risk Forum of the Bank of England (a Private Sector Forum for Developing Best Practice) Figure B2.2.1: Mapping of Investigated Scenarios to the NGFS Classification 11 Figure 2.1: Coverage of Physical Climate-Related Risks in Current Scenarios 12 Figure B2.3.1: United Kingdom Windstorm scenario 15 Figure B2.3.2: United Kingdom flood scenario 15 Figure B2.3.3: Hurricane Strike to the US Northeast Scenario 16 Figure B2.3.4: Second Hurricane Strike to South Carolina Scenario 16 Figure B2.4.1: Impact of Typhoon on Bank Capital – Normal Time 17 Figure B2.4.2: Impact of Typhoons and Pandemic on Bank Capital 17 Figure 3.1: Illustration of the Transmission Channels for 19 Shocks from the Real Economy to the Financial Sector Figure 4.1: Illustrative Statistical Simulation of Potential Tropical Storm 24 Impacts for a Highly Exposed Country Over 20 Years. Figure 4.2: Transmission Channels Between the Real 29 Economy, the Financial Sector, and the Public Sector Figure B4.1.1: Four Scenarios Considered in This Study. 30 Figure 4.3: Panels a, b, c: Three Outputs from the EIRIN Model Including a Strong Typhoon 31 Figure 4.4: Real GDP Indexed Against the BAU Scenario for a Compound Shock (Typhoon plus 32 COVID-19) in the EIRIN Model with Different Credit Constraints Figure 4.5: Compound Risk Multiplier for Two Example Middle-Income Countries, Where One Is 34 Exposed to a Flood Shock (Country A) and the Other a Typhoon Shock (Country B) During a Pandemic Figure 5.1: A Framework for Scenario Construction for Physical Climate-Related Shocks 36 1. >>> Introduction Climate change is becoming a main concern of ministries of finance, central banks, and finan- cial regulators. More than 50 ministries of finance have endorsed the Helsinki Principles, which include commitments to take action to account for climate change within macroeconomic policy, fiscal planning, budgeting, and public investment (Principles 4 and 51). To date, 100 central banks and financial regulators have become members of the Network of Central Banks and Su- pervisors for Greening the Financial System (NGFS), with its central goal to contribute to climate and environmental risk management in the financial sector. Managing the systemic risks for fi- nancial stability is a core part of the mandate of central banks. Financial regulators play a central role in assessing and managing idiosyncratic risks as well as ensuring the development of sound financial markets for the long term. A growing chorus of central banks and financial regulators have highlighted the potential financial risks associated with climate change and many, including the United Kingdom, the European Central Bank (ECB), France, Singapore, Australia, and the Netherlands, are beginning to put in place supervisory guidance and/or requirements for banks and insurers to disclose, assess, and embed climate risks within risk management frameworks. Much of the focus to date on climate-related financial risks has been on so-called climate tran- sition risks, that is, financial risks associated with the way policies, regulations, changing sen- timents, or technological shocks are introduced in the low-carbon transition (Carney 2015). Several central banks and financial regulators have started to assess investors’ exposure to transition risks via Climate Policy Relevant Sectors (Battiston et al. 2020, EBA 2020). With re- gard to climate risk exposure, a growing number of central banks have developed climate stress tests (Vermuelen et al. 2019, Allen et al. 2020, de Guindos 2021) that translate climate scenarios developed by Integrated Assessment Models (IAM) into financial risk metrics, building on the climate stress test approach developed by Battiston et al. (2017). There has been less focus on physical climate-related financial risks, although recent supervi- sory statements by central banks place equal emphasis on physical risks (see, for example, BoE 2020). This paper focuses on approaches to assess financial risks from physical climate shocks for central banks and supervisors. Physical risks arise from the changes in weather and climate that impact economies and the financial sector (FSB 2020). Activity in this area is now beginning to ramp up with guidance, scenarios, and analyses becoming available (for example, UNEPFI 2020; Smith 2021; IMF 2021) as well as new research on the financial stability implications of physical climate risks. For example, Mandel et al. (2021) find that in high-end climate scenarios, 1. For more information on the Helsinki Principles, visit The Coalition of Finance Ministers for Climate Action at https://www.financeministersforclimate.org/. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 7 physical climate risks related to flooding could lead to finan- macroprudential risk assessments as well as potentially mis- cial impacts that become commensurate with the capital of the pricing of these risks within wider financial decision making. major global banking sectors. We refer here to those scenarios made available in the pub- lic domain by, for example, the NGFS2, specifically to support Under the NGFS framework (NGFS 2020b), physical climate central banks, financial regulators, and financial institutions in risks are subdivided into two categories: chronic risks and climate stress testing and scenario analyses. This paper aims acute risks. Chronic risks result from gradual shifts in bio- to suggest how these gaps could be filled within subsequent physical and climate characteristics over time due to climate versions of the scenarios as well as those tailored scenarios change. This includes, for example, changes in labor produc- produced by central banks and financial institutions, including tivity due to gradually warming temperatures or reductions in the Bank of England 2021 Climate Biennial Exploratory Sce- agricultural output due to shifting rainfall patterns. Acute risks nario (CBES). This paper draws upon existing evidence, tools, refer to changing frequencies or severity of shocks, such as and experience from other related sectors such as insurance. natural catastrophes, including flooding, tropical cyclones, It proposes a framework for scenario generation for physical wildfire, heat waves or droughts (IPCC 2012). The NGFS climate-related financial risks to the banking sector to help fill framework and its scenarios have provided a basis for central these gaps and give an order of magnitude of the potential banks and other financial institutions to begin climate risk as- underestimate of physical climate risk. sessment exercises. This paper aims to inform global and national discussions on This paper focuses on acute risks, hereafter referred to as scenarios for acute physical climate financial risk assessment. physical climate shocks. These sudden and severe shocks, The primary goal is to inform ministries of finance, central as opposed to more long-term, gradual shifts in climate, are banks, financial regulators, and financial institutions involved most likely to generate material shocks to the financial sector in climate financial risk assessments, both for micro- and mac- in the near-term (see Feyen et al. 2020, and Calice and Miguel roprudential risk management. This includes climate stress 2021 for examples). Yet, to date, physical climate shocks have testing applications and broader scenario analysis. It also has not been explicitly considered within the core NGFS scenarios applications for financial institutions and investors using sce- (NGFS 2021b), though they are beginning to be incorporated narios to incorporate climate risks into wider financial decision in some climate stress testing by central banks (see, for ex- making, disclosures, and risk management. ample, de Guindos 2021), the World Bank, and the Interna- tional Monetary Fund (IMF) (see IMF 2021 for an example). The starting place for understanding future financial risks from Assessing the financial impacts of physical climate shocks is physical climate shocks is to first assess how such shocks nontrivial and requires drawing upon expertise from across affected the financial sector (in particular, the banking sector multiple disciplines, including climate science, earth sciences, but also the insurance industry) in the past and to fully map engineering, economics, and finance. There is a deep litera- potential transmission channels that could come into play in ture and practice on assessing the economic impacts of physi- the future. As such, this paper incorporates a review of the cal climate shocks that can be drawn upon. empirical evidence on the impacts physical climate shocks on the financial sector and financial stability. A similar approach This paper argues that there are gaps in the way acute risks is proposed by the Climate Financial Risk Forum of the Bank are included within current scenarios for physical climate-re- of England (CFRF 2020) for individual financial institutions lated financial risk assessments. This could lead to underes- as starting point to generating relevant climate scenarios for timating the potential scale of those risks within micro- and stress testing (see figure 1.1). 2. Climate scenarios are available at https://www.ngfs.net/ngfs-scenarios-portal/. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 8 > > > F I G U R E 1 . 1 . Approach to Scenario Generation and Analysis Proposed by the Climate Financial Risk Forum of the Bank of England (a Private Sector Forum for Developing Best Practice) End to end Climate Scenario Analysis Process 3. Condut exposure analysis Identify potential 2 exposures to 1. climate-related 5. 6. Identify Climate- Examine risks Technological Climate policy related Transmission evolution landscape risks channels Scenario Develop suitable Analysis climate-related scenarios Process 4. 7. Emission and Socio-economic temperature context 10. pathways 8. Assess Financial Assess the Define Risk impacts and take financial measure appropriate action impact 9. Choose Impact assessment tools Source: CFRF 2020 The paper is organized as follows. Section 2 begins with a risk assessment and proposes a framework for scenario gen- review of current scenarios provided in the public domain for eration based upon this. This section then presents evidence climate financial risk assessment by the NGFS and others and on the scale of the implications for climate-related financial compares these with the wider literature on the economics of risks. Estimates of scale are provided by reviewing current climate risks and scenario development for stress testing by fi- literature and by presenting a case study for a generic mid- nancial institutions. This section identifies the main gap versus dle-income country that is highly exposed to extreme weather current understanding of physical risks and practice in other events (for example, typhoons and flooding) based on new analogous areas. Section 3 then reviews the current empirical analyses using the EIRIN model (Monasterolo and Raberto evidence on the economic and financial impacts of physical 2018; Dunz et al. 2021). Section 5 discusses next steps to- climate shocks on the banking sector to identify five specific ward implementation of this framework. The paper concludes gaps. Section 4 discusses how these gaps can be better ad- by drawing recommendations for future research to fill the pri- dressed within scenarios for physical climate-related financial ority data gaps identified in this study. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 9 2. >>> Review of Current Scenarios for Physical Climate Risks To frame the context, we first review the scenario generation for physical climate-related risks as embodied by the Network of Central Banks and Supervisors for Greening the Financial System (NGFS) (NGFS 2020a; NGFS 2020b; NGFS 2021b). Box 2.1 introduces the six policy scenarios defined by the NGFS. Physical risk scenarios use climate projections from five Global Climate Models (GCMs) driven by the Representative Concentration Pathways (RCPs). The NGFS recommends the use of the climate impact scenarios collected by the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP) initiative3 to carry out estimates of climate physical risk for investors. These impact scenarios cover climate impacts such as changes in agricultural productivity, ecosystems, forestry, and water stress. For economic damages, NGFS climate physical risk scenarios (identified as “damage” scenarios) rely upon damage functions developed in other models (for example, William D. Nordhaus’ Dynamic Integrated Climate Economy (DICE) model; Kalkuhl and Wenz 2020). These scenarios are then embedded in process-based Integrated Assessment Models (IAMs)4 (that is, GCAM, MESSAGE-GLOBIOM, REMIND-MagPie) to provide an estimate of global, aggregate gross domestic product (GDP) loss. Damage functions link climate variables, such as mean temperature, with impact metrics, such as GDP losses based upon econometric analyses and other evidence. The resulting trajectories can be calculated by users in the NGFS scenario explorer5. These show that estimates of physical climate losses are very sensitive to the type of damage function and its calibration. 3. To learn more about the ISIMIP initiative, visit https://www.isimip.org/outputdata/. 4. A more recent generation of “process based,” large-scale IAM, embeds a granular representation of energy technologies (for example, fossil fuel and renewables) (Weyant 2017). They develop long-term emission projections and socioeconomic scenarios, based on assumptions on carbon pricing and modeling of technology investments that suggest how to reach given targets in terms of cumulative emissions (and thus in terms of carbon budget) by 2100. Emissions translate then into temperature targets with associated probabilities. Process-based IAM do not directly model disaster risk yet focus on the transition to low-carbon futures. 5. See https://data.ene.iiasa.ac.at/ngfs to learn more about the NGFS Scenario Explorer. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 10 > > > B OX 2 .1 Overview of NGFS Scenario Framework The NGFS has recommended a set of climate scenarios to be used by financial supervisors for climate-related financial risk assessment (including stress test exercises, for example, see guidance document NGFS 2020a); these aim to inform assessments of both transition and physical climate-related risks. These scenarios are based upon process-based Integrated Assessment Models (IAM) that are reviewed by the Intergovernmental Panel on Climate Change (IPCC) (IPCC 2014; IPCC 2018) and have a high-granular representation of energy technologies to aid in transition risk assessment. Within the dimension of (high) physical risk, the NGFS has identified two high-level scenarios (NGFS 2021b): 1. Hot house world corresponds to the IPCC scenario Representative Concentration Pathways (RCP) 6.0—a scenario in which the global temperature reaches over 3 degrees C by 2100 in comparison with pre-industrial times. This is described as a situation in which output across low/high carbon activities progresses in line with the current NGFS scenarios policies, that is, a continued reliance on fossil fuel energy sources and unabated greenhouse gas emissions. Along this pathway, an increase of physical risk is projected due to increased frequency and intensity of climate-related extreme events (or physical climate shocks) and chronic effects (such as sea level rise and permafrost melting). In this scenario, there is no transition to a low-carbon economy, and hence there is no transition risk. 2. Transition scenarios (orderly and disorderly) (figure B2.1.1) corresponds to scenario RCP2.6—below 2 degrees Celsius (C) by 2100. These scenarios are further subdivided into subscenarios: one considering different temperature targets (1.5 degrees C or 2 degrees C. respectively); another considering the timing of the introduction of climate policy such as a carbon tax (immediate, meaning 2020 versus delayed, say, to 2030); and finally, reliance on carbon dioxide removal.a > > > F I G U R E B 2 . 1 . 1 . - Mapping of Investigated Scenarios to the NGFS Classification HIGH Disorderly Too little, too late Divergent net-zero (1.5oC) Delayed (2oC) Transition Risks Net-zero 2050 (1.5oC) Bellow (2oC) NDCs Current policies Orderly Hot house world LOW LOW Physical Risks HIGH a. Arguably, the assumed mitigation pathway of the transition scenarios may also be optimistic, for example in assuming the full deployment potential of carbon capture and removal technologies (such as from geoengineering, afforestation, soil and water management, etc.) and assuming that countries are on a track on their 2030 policy commitments (including nationally determined contributions). According to the last United Nations Environment Programme (2021) Emissions Gap Report, most countries who reported their progress are still far from their goals. However, in this paper we focus on the quantification of impacts rather than the mitigation paths. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 11 Challenges to using IAMs for this type of assessment are well by current scenarios. Each of these is important to consider documented (Farmer et al. 2015; Stern 2016; Hepburn and when structuring scenarios for the purpose of evaluating the Farmer 2020). Firstly, such models present an incomplete risk of financial outcomes and potential risk management picture of the impacts of climate change, including missing options. Figure 2.1 represents the coverage of risks and extreme weather shocks (Stern 2016). In addition, several uncertainties in current scenarios based on the authors’ phenomena induced by climate change—such as migrations, analysis, using a framework adapted from Watkiss et al. 2005 crop yield shocks, and social instabilities in exposed regions— and Stern 2006. are missing from these models. The potential for cascading and compounding risks or nonlinear effects are also missing The scenarios also do not capture the potential policy and (Hepburn and Farmer 2020). The links between climate and financial responses to changing physical risks—for example, ecosystems and natural resources (such as soil, water, for- the potential for rapid adjustments in asset valuations in coastal estry) that are known to be an important driver of financial risk and inland flood-exposed regions as investors perceive growing (Dasgupta 2021), are excluded. Taken together, this implies risks related to climate change or shifts in public policy that have that current IAMs and scenarios built upon them could under- widespread impacts on the availability of insurance. Such rapid estimate physical climate risks. Finally, and crucial for climate adjustments are not unheard of today (for example, Keys and financial risk assessment, IAM scenarios do not account for Mulder 2020; Kyum Kim and Peiser 2020). the financial sector and investors’ expectations, thus missing important feedback between the economy and financial sec- NGFS made available in 2021 a Climate Impact Explorer,6 tor (Battiston et al. 2021). Evidence suggests these could be with a set of indicators of acute risks such as 1-in-100-year substantial gaps in current physical risk scenarios. losses and population exposures to extremes (from Climate Analytics). The Bank of England released similar variables for There are also limitations in current climate and impact models its 2021 Climate Biennial Exploratory Scenario (CBES) (Box that underpin IAMs and the representation of uncertainties, as 2.2).7 This is helpful in filling the gaps in the core scenarios, outlined by Fiedler et al. (2021) and Farmer et al. (2015): the yet this does not fully fill the gaps identified by Fiedler et al. range of possible future outcomes is much wider than implied (2021), Hepburn and Farmer (2020) and Stern (2016). > > > F I G U R E 2 . 1 . - Coverage of Physical Climate-Related Risks in Current Scenarios Scope of Impacts included in Financial Risk Scenarios Chornic Climate Physical Climate Shocks Complex feedbacks, Change and Impacts Direct Impacts Indirect Impacts climate-nature, real economy-financial sector Scenario-based (multi-climate deterministic Limit of Representation of Uncertainties model) coverage of most IAMs, as NGFS Climate well as NGFS Impact Explorer 2021 and BoE 2021 CBES Scenarios None None NGFS-ISIMIP Data Portal Uncertainties (incomplete sampling) Bounded None None Surprises Change/ System None Source: Adapted for this publication from diagram for coverage of risks in IAMs in Watkiss, Downing et al. 2005 and elaborated in Stern 2006. NGFS 2021 = Pub- lished scenarios in June 2021 (NGFS 2021b). NGFS-ISIMIP= Data available through the NGFS Scenario Explorer (https://data.ene.iiasa.ac.at/ngfs). BoE 2021 CBES are as outlined in box 2.2. NGFS Climate Impact Explorer (http://climate-impact-explorer.climateanalytics.org/). 6. More information is available at http://climate-impact-explorer.climateanalytics.org/. 7. Key elements of the 2021 biennial exploratory scenario are found at https://www.bankofengland.co.uk/stress-testing/2021/key-elements-2021-biennial-exploratory-scenar- io-financial-risks-climate-change. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 12 According to the International Monetary Fund (IMF), for stress impact” (IMF 2019). The evidence above suggests that the po- testing, the priority for central banks and supervisors is to tential for systemic risks is currently not fully captured by the identify and assess macrofinancial vulnerabilities that can current scenarios made available to central banks and finan- trigger systemic risk, or, through the operation of the finan- cial institutions. Using scenarios that do not capture important cial system, create downside risks to growth and so signal the drivers of material financial risk could constitute a source of need of systemwide mitigating measures (IMF 2019). Scenar- future systemic risk. For example, the emergence of new risk ios for bank stress testing should be “forward-looking, severe, information over time could lead to rapid devaluations in asset consistent, and robust trajectories for a comprehensive set of prices and knock-on effects that could create instability. The macro-financial variables that react following the materializa- same argument can be made for stress testing idiosyncratic tion of shocks… Scenario design starts with a narrative about risks (microprudential regulation) and ensuring the long-term how the realization of tail risks could interact with financial vul- soundness of financial sector development (often the respon- nerabilities to generate severe but plausible macro-financial sibility of financial regulators and ministries of finance). > > > BOX 2.2 Acute risks in the Bank of England 2021 Climate Biennial Exploratory Scenario The Bank of England (BoE) runs regular stress tests to help assess the resilience of the United Kingdom’s financial system and individual institutions. There are two types of exercise within the BoE’s concurrent stress testing framework for banks and building societies (hereafter ‘banks’): annual solvency stress tests and biennial exploratory scenarios. Running biennial exploratory scenarios allows policymakers to probe the resilience of the United Kingdom financial system to a wide range of risks and is a tool to enhance participants’ strategic thinking on how to manage those risks. The 2021 exercise explores the resilience of the largest United Kingdom banks and insurers to the physical and transition risks associated with climate change, including acute physical risks. For the BoE, the intention is that the Climate Biennial Exploratory Scenario (CBES) be a learning exercise. Given that expertise in modeling such risks is in its infancy, the exercise aims to develop the capabilities of both the BoE and CBES participants. The scenarios provided by the BoE for the CBES are not forecasts of the most likely future outcomes. Instead, the BoE describes that scenarios are plausible representations of what might happen based on different future paths of governments’ climate policies (policies aimed at limiting the rise in global temperature). Each scenario is assumed to take place over the period 2021–2050. Participants will measure the impact of the scenarios on their end-2020 balance sheets, which represents a proxy for their current business models. For banks, the CBES focuses on the credit risk associated with the banking book, with an emphasis on detailed analysis of risks to large corporate counterparties. A key metric of that risk will be the cumulative total of provisions against credit-impaired loans at various points in the scenarios. Chronic physical risk variablesa are provided at (mainly) the national level for the United Kingdom and seven other countries, including temperature; precipitation rate; wind speed; land area exposed to crop failure and sea level rise in 2020, 2030, and 2050. Acute risk variables include tropical cyclone (category 3–5 frequency and intensity changes) and land area exposed to heat wave and wildfire. Macroeconomic variables are provided, including both transition and (chronic) physical risks. Much of the data comes from the NGFS, with some exceptions.b A single value is given for each variable and no uncertainty information is provided. The BoE provides links to additional data provided by the NGFS from the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP)c that includes further model-based projections. Source: Bank of England 2021, Key elements of the 2021 biennial exploratory scenario located at https://www.bankofengland.co.uk/stress-testing/2021/key- elements-2021-biennial-exploratory-scenario-financial-risks-climate-change. a. Chronic variables can be found at https://www.bankofengland.co.uk/-/media/boe/files/stress-testing/2021/variable-paths. b. OASIS Hub for wind speeds and sea level rise, UK Met Office for UK projections and Knutson et al. 2000 for tropical cyclone. c. The physical impact data collected by the ISIMIP is located at https://www.isimip.org/outputdata/isimip-data-on-the-esgf-server/. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 13 This paper discusses how such gaps can be filled in a way that enough capital must be held to cover the market-consis- that is feasible and commensurate with the scale of the risks tent losses that may occur over the next year with a confi- versus other (non-climate) risks faced by financial institutions. dence level of 99.5 percent.8 The Bank of England’s General It is helpful to briefly compare and contrast the approach to Insurance Stress Test 2019 required insurers to stress test scenario generation taken to standard practice in macro- and against extreme weather event scenarios of between 1-in-100 microprudential risk management in the insurance sector (in- years and 1-in-250 years compounded with an insurance as- cluding regulatory frameworks such as Solvency II—the pru- set price shock (PRA 2019). Lloyd’s of London requires all dential framework for insurance firms in the European Union). syndicates to report against 16 compulsory “Realistic Disas- The insurance industry and its supervisors and regulators are ter Scenarios,” including a major hurricane striking New York experienced in managing the financial risks of physical climate State and the East Coast of the United States (Lloyd’s of Lon- shocks. The insurance sector typically uses catastrophe risk don 2021). See Box 2.3 for more on these two approaches. models that are tailored to assess the direct impacts of physi- These approaches can draw important lessons for physical cal climate shocks (physical damage) and importantly, are climate-related financial risk assessment; specifically, the fo- able to represent the volatility (or stochastic nature) of these cus on simple, realistic but extreme scenarios for stress test- shocks as well as their correlation/systemic implications for ing and the use of multiple scenarios, with quantitative and individual firms and the industry globally, rather than just aver- qualitative elements, that aim to explore the space of possible ages. Such extreme scenarios are critical to inform underwrit- outcomes and avoid spurious accuracy that can emerge when ing and portfolio risk management. Under Solvency II, capital attempting to provide projections based on models when there requirements are determined on the basis of a 99.5 percent is deep uncertainty. (1-in-200 years) value-at-risk measure over 1 year, meaning 8. Solvency II (Directive 2009/138/EC) as amended by Directive 2014/51/EU (Omnibus II). EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 14 > > > BOX 2.3 Stress Testing Scenarios for Physical Climate Shocks in the Insurance Industry Bank of England’s General Insurance Stress Test 2019a The 2019 stress test provided a series of scenarios that aimed to stress the asset and liability side of insurers in parallel. On the asset side, the scenario outlines a deterioration in the economic environment, including reduction in interest rates, widening of corporate bond spreads, and fall in asset values. In parallel, insurers are asked to stress test against five liability shock scenarios, four of which are based on natural catastrophes. As an example, scenario five describes a set of two events that generate some £20 billion of aggregate insured loss, both occurring in the United Kingdom. The first event is a windstorm causing significant storm surge losses along the East coast of England that generates approximate half of the overall losses (figure B2.3.1). The second event is for extensive flooding across England and Wales, generating the remainder of the overall losses (figure B2.3.2). The return period for aggregate wind, surge, and flood losses of this size to the United Kingdom is estimated to be approximately 200 to 250 years. Firms are encouraged to develop their own view of risk, including making adjustments for model uncertainty. This stress is superimposed on the insurance asset shock scenario, a compound risk scenario. In 2019, firms were also requested to consider the expected impact under three different climatic states on their assets, liabilities, and business models, assuming that their current exposures and investment profile remain constant. Learning from this exercise fed into the design of the 2021 Climate Biennial Exploratory Scenario. F I G U R E B 2 . 3 .1 . FIGURE B2.3.2. United Kingdom Windstorm scenario United Kingdom flood scenario ... EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 15 Lloyd’s of London Realistic Disaster Scenarios 2021b There are 16 compulsory scenarios that managing agents must complete for all syndicates, with losses of up to US$120 billion. Each scenario outlines an extreme but realistic event, the largest of which is a double hurricane strike to the United States in a year: one causing major damage to the North East United States and the other to South Carolina. Managing agents are provided with detailed loss information for the scenarios, including residential and commercial property losses as well as disruptions to ports and airports. Other scenarios include windstorm strikes to Florida, a United Kingdom flood, Japanese typhoon and earthquakes, and a California earthquake and terrorism scenarios. FIGURE B2.3.3. FIGURE B2.3.4. Hurricane Strike to the US Northeast Scenario Second Hurricane Strike to South Carolina Scenario Source: PRA 2021; Lloyd’s of London 2021 a. Information resourced at https://www.bankofengland.co.uk/-/media/boe/files/prudential-regulation/letter/2019/general-insurance-stress-test-2019-scenario-specifica- tion-guidelines-and-instructions.pdf. b. Information resourced at https://assets.lloyds.com/media/e73cc2f7-a535-4eaf-8196-fedcf5e1432c/2%20RDS%20Scenario%20Specification%20%20January%202021.pdf. One example of a central bank climate stress test that bridged tastrophe risk model, and a macrofinancial model to develop across this standard practice in the insurance industry and scenarios for the stress test. It concluded that physical climate approaches for bank stress testing was the 2021 IMF and risks are relevant for financial stability, though the infrastruc- World Bank Financial Sector Assessment Program for the ture destruction from typhoon wind alone is not systemic un- Philippines (IMF 2021). Similarly, to the 2019 Bank of England less extreme tail events materialize. This highlights the im- General Insurance Stress Test, this climate stress test for the portance of considering such tail risks. On the basis of this central bank considered a 1-in-250 years typhoon scenario analysis, it recommended improving information collection, compounded with an economic shock and a pandemic, in a monitoring risk metrics, and stress test capacity for climate current and future climate (taking an upper bound scenario). change and environmental risks. The climate stress (Box 2.4) connected a climate model, a ca- EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 16 > > > BOX 2.4 IMF 2021 Philippines Financial System Stability Assessment Program The 2021 Financial Sector Assessment Program (FSAP) developed a new approach for analyzing banks’ solvency for physical risks from typhoons, building climate change macroeconomic scenarios using climate science studies, a catastrophe risk model, and a macrofinancial model, in collaboration with the World Bank. The analysis indicated the relevance of typhoon risks, though found that they may not be necessarily systemic except for extreme tail events. Without other shocks, the destruction of physical capital from typhoons’ wind alone would reduce bank capital ratio only by 1 percentage point, even in once-in-500-years events in the future (figure B2.4.1). However, the joint shock with a pandemic intensifies the effects of climate change for extremely intense typhoons (figure B2.4.2). For once-in-500- years events, the difference between current and future scenarios with the pandemic rises to 4½ percentage points. F I G U R E B 2 . 4 .1 . FIGURE B2.4.2. Impact of Typhoon on Bank Capital – Impact of Typhoons and Pandemic Normal Time on Bank Capital (Total capital adequacy ratio in percent) (Total capital adequacy ratio in percent) 18.0 18.0 Typhoon, Typhoon, 16.0 once in 25 16.0 15.3 once in 25 15.3 years years October WEO 14.0 14.0 12.0 12.0 January WEO Typhoon, 10.0 once in 500 10.0 9.7 10.3 10.2 years 8.0 8.0 9.1 8.9 9.6 7.7 6.0 6.0 6.8 4.0 4.0 5.4 5.2 Typhoon, 2.0 2.0 once in 500 3.2 years 1.0 0.0 0.0 2019 2020 2021 2022 t t+1 t+2 t+3 2019 2020 2021 2022 t t+1 t+2 t+3 Current scenario Future scenario Current scenario Future scenario Note: WEO = World Economic Outlook, https://www.imf.org/en/Publications/WEO EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 17 3. >>> Empirical Evidence on the Impacts of Physical Climate Shocks to Banks This section reviews the empirical literature on the impacts of physical climate shocks on the banking sector in order to identify specific gaps in the representation of risks and transmission channels. While there is a well-established literature on the social and economic impacts of weather-related shocks based on empirical analyses and models (for example, review by Botzen, Deschenes, and Saunder 2019), evidence on the impacts on the banking sector itself is more nascent. A number of studies have analyzed historical impacts of disaster-related shocks on the banking sector. For example, Noth and Schüwer (2017) present evidence from the USA that disasters can weaken the stability of banks in the same region measured through lower z-scores, higher probabilities of default and higher nonperforming loan ratios. Klomp et al. (2014) present evidence across 140 countries showing increased probabilities of default for commercial banks following disasters. Calice and Miguel (2021) study empirical evidence on climate risks in Latin America and the Caribbean (LAC) and find that after largescale natural disasters, banks’ nonperforming loans increase by up to 1.4 percentage points in affected provinces. They conclude that in terms of physical climate risks, exposure to floods represents the most important source of credit risk for the LAC banking sector, with exposure particularly concentrated around cities. Climate can also affect bank lending decisions; for example, Garbarino and Guin (2021) analyze how lenders account for recent severe flood events in England in 2013–14, finding that lender valuations are biased upward, and lender do not track closely the impact of extreme weather events. A first conclusion from such studies is that the impact of physical climate risks on the banking sector is highly dependent on the level of resilience of the financial sector overall (to any shock) and the vulnerability of their borrowers. For example, countries with weaker supervision and regulation and with more concentrated and less interconnected banking sectors will see greater risks, while more advanced and resilient financial sectors will be less affected. Smaller economies will be more vulnerable, particularly small island states where economic losses can constitute a significant proportion of their gross domestic product (GDP). This is an intuitive result, yet one that is important to explicitly recognize because it underlines the importance of tailoring scenarios to the circumstances of the country. This moderating (or amplifying) factor needs to be accounted for when assessing climate-related financial risks and is not explicitly EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 18 captured in current scenarios. Central banks and supervisions example, due to disruptions to power) and reduced demand would typically account for this within their own top-down or reduced production (for example, in the case of agricultural stress tests, but there is limited guidance on the implications firms and drought), potentially leading to reduced return for climate scenario construction or analysis. on assets for the firm and reduced asset quality for banks (higher nonperforming loans or impairments). This can have The evidence available also points toward a complex web knock-on effects throughout the real economy that can slow of transmission channels between the direct impacts of growth, including through impacts on supply chains, demand, shocks (such as capital destruction or loss utilities such as and households (impacts on income and consumption). At power), the real economy, and the financial sector. Figure the same time, demand for credit increases postdisaster for 3.1 below maps the main transmission pathways based on recovery and reconstruction, and in some cases, withdrawals the literature and describes the main types of impact in the may increase, tightening liquidity of banks. context of the International Monetary Fund’s (IMF’s) Financial Soundness Indicators (IMF 2006) (capital adequacy, asset Where financial resilience of banks is relatively low, this can quality, earnings and profitability, liquidity, and sensitivity to lead to a depletion of capital, and, if liquidity is insufficient, can market risk). It is consistent with that provided by the Bank for threaten the survival of the bank. Reserves may be depleted International Settlements (BIS) (BIS 2021) and the Network due to a large write-off of loan losses or a reduction in risk- of Central Banks and Supervisors for Greening the Financial weighted asset values. Asset risk may be increased by the System (NGFS) (NGFS 2021a) as well as the analysis of destruction of collateral of borrowers. Profitability (return on transmission channels provided by Feyen et al. 2020. assets) may decrease as a result of write-offs. Banks may also experience damage and operational disruption from physical Disasters cause damage to physical assets and other damages to their own buildings or critical services. productive capital, as well as business interruption (for > > > F I G U R E 3 . 1 . - Illustration of the Transmission Channels for Shocks from the Real Economy to the Financial Sector Disaster Impacts from the real economy Potential impacts within the financial sector Productivity Withdrawal of Reduced Increased loan Impacts bank’s deposit income/revenue delinquency and (including (Liquidity) business foreclosures Depletion of interruptions) capital levels Reduced value (Capital adequacy) Reduced of collateral asset values Damages (Asset quality) Damaged bank Reduced profit to Capital operations levels (Earnings) (including (Management) physical assets) Demand for Demand for Higher lending Inter-bank lending recovery and financing from local bank (Liquidity) reconstruction Inter-bank lending (Liquidity) Higher lending Reduced trade Liquidity impacts from non-local/ and investment on another bank Contagion into foreign bank another country Source: Original figure for this publication. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 19 Where predisaster financial resilience and health of the banking 4.5 percent growth in 2013 mainly due to the lingering effects sector is high, the impact of such shocks on the financial of the typhoon (Perez 2015). In Nepal, while overall financial sector is minimal. In addition, even in developing countries, sector stability was unaffected by the 2015 earthquake due to typically the government and central bank will act quickly swift action by the NRB, many microfinance institutions (MFIs) postdisaster to protect stability. For example, to help ease the and savings and credit cooperatives (SACCOs) serving more impact of the 2015 earthquake, the central bank, Nepal Rastra deprived areas suffered increases in NPLs and liquidity issues Bank (NRB), put in place regulatory relief for banks to enable (Government of Nepal 2015). This suggests that subsectors them to continue to extend credit despite heavy impacts on serving more vulnerable groups, such as agribanks, MFIs, the economy, including time-bounded measures covering and SACCOs may face much higher climate-related financial loan-loss provisioning, loan rescheduling, grace periods, and risks. More work is required to understand any potential regulatory forbearance on asset classification (IMF 2015). contagion affects to financial stability overall as well as the Impacts on financial stability were avoided. Likewise, in the implications for financial inclusion, economic development, 2011 Thailand floods, 4–13 percent of loans by value were and poverty alleviation. in affected areas, and the Bank of Thailand responded by relaxing asset classifications (TCG 2016; World Bank 2012; The indirect impacts of physical climate shocks on the banking IMF 2012; Ramcharram 2017) and providing a (partial) sector through transmission through the wider macroeconomy, credit guarantee facility to support recovery. This, combined could be larger than the financial risks associated with the with strong existing capital buffers, avoided major impacts. direct impacts on firms. This effect is also highly dependent These responses are one reason why it is difficult to find on economywide vulnerability factors. For example, for the empirical evidence of impacts of disasters on, for example, most high-income countries, the overall economic impact of nonperforming loans. disasters is typically small compared with GDP, and indeed, reconstruction can boost output, creating increased demand Arguably, this risk is being absorbed somewhere in the system, for credit. But for small island developing states (SIDS) and and this should be considered by ministries of finance and other small or highly vulnerable states, the impacts of climate central banks when projecting future climate-related financial shocks on the economy can be significant and long-lasting. risks. In some cases, for example, the risk is being absorbed For example, in the Caribbean, a one-off hurricane strike by the public sector in the form of partial credit guarantee can cause damages equivalent to more than 100 percent of schemes and so is relevant to fiscal risk assessments. GDP, creating, on average, a reduction in output growth of just under 1 percent since 1950 or 7.6 percent for the most Feyen et al. (2020) demonstrate that countries with the destructive hurricanes (Stobl 2009; Alleyne et al. 2017). The greatest physical climate risks also tend to be those with 2019 Financial System Stability Assessment Program for the the greatest macrofinancial risks. They find that a significant Bahamas found a relationship between NPLs and hurricanes number of countries, particularly emerging and developing and concluded that the most significant impacts of hurricanes countries, face a double jeopardy due to the simultaneous on the banking sector are mediated through the impacts of presence of climate-related and macrofinancial risks. These hurricanes on economic growth and employment rather than countries have limited macrofinancial capacity to act, meaning direct credit exposure (IMF 2019). that as physical climate risks materialize, high macrofinancial risks mean low macrofinancial resilience and a high risk of Physical climate-related financial risks become more material prolonged crisis. when compounded with other economic shocks. For example, evidence from the Caribbean showed that when growth is weak, Even where national (systemic) financial stability may be the impact of hurricanes on NPLs can be amplified and more minimally affected, the impacts on local banks and financial nonlinear (IMF 2019; Brei, Mohan, and Strobl 2019). This is institutions serving more vulnerable affected groups can be an important finding, particularly in a COVID-19 context when significant. For example, while Typhoon Haiyan (Yolanda) in economies are under strain. For instance, the compounding of the Philippines in 2013 had a minimal and short-lived impact physical climate risk (hurricanes) and the pandemic in Mexico on national GDP, the impact on the local economy and local contributed to amplify the initial macroeconomic shock, with banks was significant (Gonzalez Pelaez 2019); in the worst-hit implications for banks’ financial stability and sovereign debt areas of Leyte Province, damage to the main sugar cane and sustainability (Dunz et al. 2021). This points to the importance rice industries was estimated at over US$300 million (World of considering compounding risks within a physical climate Bank Group 2014), and the wider Eastern Visayas region risk assessment (Ranger, Reeder, and Lowe 2021). overall suffered 2.3 percent contraction in GDP in 2014 from a EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 20 Finally, empirical evidence shows that feedback effects of finance, central banks, financial regulators, and financial between the financial sector and the real economy can institutions in physical climate financial risk assessment for amplify or dampen the financial risks to banks (and the climate-related financial risk management, including stress wider economy). Firstly, banking-related services play a testing. Those factors include: critical role in economic recovery postdisaster; this includes • The importance of representing the feedback between the withdrawals of deposits, restructuring lending, new lending real economy and the financial sector, which will be spe- to finance postdisaster reconstruction, or remittance flows. cific to the circumstances of the country. Any disruption to these services can have a major impact on • The need for a tailored approach to developing scenarios the scale of impacts on firms and households and recovery that fully capture the risk transmission channels of great- times. For example, evidence from 1995 Kobe and 2011 est relevance to the financial sector. Tohoku earthquakes in Japan suggests that physical damage • The links between public and private financial institutions; to banks negatively affected investment by firms in those risk is always absorbed somewhere and often the public regions (Miyakawa and Hosono 2017). Empirical analyses sector absorbs some of the private financial risk in times over 178 countries from 1979 to 2007 found that lack of of crises. credit can compound the effects of a disaster, and countries • The importance of representing the full economic and so- with lower financial sector development tend to suffer more cial impacts of physical climate shocks, including extreme persistent negative impacts of disasters on economic growth events and their short-term and long-term indirect eco- over the medium term (McDermott, Barry, and Tol 2014). This nomic impacts. feedback could create greater vulnerability over time as asset • The need to map subnational risks and risks to financial quality increasingly erodes, particularly in the context of more institutions serving the most vulnerable groups to under- frequent, intense climate-related shocks. stand potential contagion effects (and impacts on finan- cial inclusion). In conclusion, the review of the empirical evidence suggests • Consideration of global cascading and national com- a number of factors that should be considered by ministries pounding risks. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 21 4. >>> Revisiting the Development of Physical Climate Financial Risk Scenarios From the review in the sections above, we can draw out five important areas to consider within a physical climate-related financial risk assessment and scenario design. 1. 2. 3. 4. Representing the current and future risks of climate extremes, such as hurricanes, droughts, and floods—or disaster scenarios—in the analysis. This includes representing the direct impacts of extreme weather on natural and human systems, for example on agricultural production, critical infrastructure services, or ecosystem services. Fully accounting for uncertainties in climate and impact models to ensure that scenarios span the space of potential future climate outcomes (rather than model averages). Including compounding scenarios of physical climate shocks with other shocks and stressors. Climate change will not happen in isolation, and physical climate risk assessments cannot ignore the compounding impacts with other factors that amplify risks, including economic cycles and socioeconomic vulnerabilities, as well as the potential for climate to compound with other shocks like pandemics and climate transition risks. Representing the indirect impacts of weather extremes on households and firms and the macroeconomic impacts in addition to the direct impacts in terms of physical capital destruction or production loss. These can create an amplification factor on the risk to the banking sector. This could also include indirect effects related to regional or global impacts of weather extremes. 5. Representing the financial sector adequately to capture both the level of resilience of the financial sector to shocks and the complex feedbacks that can amplify risks by prolonging reconstruction and recovery. This includes considering more vulnerable parts of the financial sector, such as microfinance institutions (MFIs) and savings and credit cooperatives (SACCOs) where these are material. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 22 It should be noted that in the first three areas of concern, the 4.1. Scenario Generation for Extreme exogenous shock is represented within scenarios, that is, the external driver associated with physical climate change. In the Weather Events Under Current latter two areas of concern, the representation of endogenous and Future Climate Change (internal to the economy) factors might amplify or suppress the impacts of the shock within scenarios and/or within models used to simulate the impact on the real economy and financial The 2020 and 2021 Network of Central Banks and Supervisors sector. Any standard framework for stress testing physical for Greening the Financial System (NGFS) scenarios did not climate risks should be redefined to include consideration include acute physical climate shocks, albeit in 2021 a set of these five factors. It should fully capture the nature of of separate risk indicators were provided.9 The 2021 Bank the exogenous shock and particularly the tail risks, which of England Climate Biennial Exploratory Scenario (CBES) means the volatile nature of weather extremes, the range of scenarios similarly included some scenarios of future acute possible climate outcomes, their physical impacts on natural risks, specifically for tropical cyclone. Smith (2021) reviewed systems, and the potential for compounding risks, all of which approaches taken by commercial providers and financial are known from empirical analyses to be important to banks institutions. in determining the financial risks associated with physical climate (see Section 2). A framework should also capture In the climate financial risk literature, two approaches have the interaction of these tail risks with the real economy and begun to be explored: (1) to use average annual (direct) financial vulnerabilities, which can be a critical amplifier of losses and (2) to develop a risk rating based on exposure to physical climate-related financial risks (Mandel et al. 2021). hazards10 (for example, Smith 2021). There are challenges As noted above, current scenarios do not capture consistently in both approaches for the assessment of financial risks. some of those important risk drivers. The generation of probabilistic risk data and scenarios of extreme weather to inform financial decision making is well It is necessary to take a proportionate approach. The important developed, in particular in the insurance industry. From this question is then how material are these risk drivers compared experience, the financial risks from physical climate shocks with drivers considered in the current scenarios, and how can cannot be approximated by considering only average annual these be included in scenario development in a simple way? costs of weather extremes, even on long timescales. Larger, The following subsections provide evidence on the potential rarer events can cause significant damage and disruption materiality of each of these risk factors in turn, including new and have long-lived impacts. Vulnerability to such shocks, modeling on the interaction of physical climate shocks with and in particular indirect damages, can be strongly nonlinear. financial sector vulnerabilities and the potential amplification This means that using measures of average direct losses from compound shocks. They also review the tools and over time can mask events (and tail risks) that could create approaches available to characterize the risks. While we significant damage. This is illustrated in figure 4.1. Simple take each factor in turn, the examples builds across each exposure mapping also fails to capture these nonlinear and subsection, such that the results shown for the final subsection diffuse impacts. In addition, the past is not a good guide to include each of the five factors to demonstrate how they can future risk; using only catalogs of historical events can lead be combined. to underestimates of the risk; particularly where considering the risks associated with rare, more severe shocks that may not yet have been observed in recorded history and in the context of a changing climate. These lessons have been learned at high cost by the insurance industry. For example, when Hurricane Andrew struck Florida and Louisiana in 1992 it caused (at that time) unprecedented losses that drove some insurers into insolvency as a consequence of underestimating risk within pricing and portfolio risk management. 9. Future vintages of NGFS scenarios plan to be extended to the analysis of expected economic damages of climate change, based on the probabilistic natural catastrophe impact model CLIMADA. See https://wcr.ethz.ch/research/climada.html for more on CLIMADA. 10. For example, is a particular asset located in a flood plain. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 23 > > > F I G U R E 4 .1 . - Cumulative loan losses for transition scenarios per bank (percentage of total assets) 0.050 Capital Destruction (% total private assets) 0.045 0.040 0.035 0.030 0.025 0.020 0.015 0.010 0.005 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 AVE 1 IN 10 1 IN 100 Source: Original calculations for this publication. Note: Two possible versions of a 20-year period are shown and compared with the average annual loss (each have roughly the same average annual loss). The economic and financial sector impacts associated with these three timeseries would be different given the nonlinearities in indirect impacts and the more prolonged reconstruction and recovery times associated with larger events. Disaster risk finance (DRF) analytics is a toolkit that has been to design strategies to reduce financial risks from disasters to used in practice for more than two decades to support minis- government balance sheets (World Bank 2019). tries of finance, insurance supervisors and the insurance in- dustry to strengthen their financial resilience to climate-related A further advantage of more sophisticated catastrophe risk and other shocks. Core to this toolbox is the catastrophe risk models is their ability to assess risk at a granular scale (up to model, the workhorse analytical tool of the insurance indus- individual buildings) and so assess potential concentrations try since Hurricane Andrew in 1992, to price and manage the of risk or distributional factors that can have significant impli- financial risks associated with a wide range of catastrophe cations for public policies. For banks, which may have geo- events, from pandemics to natural hazards, terrorism risks, graphically concentrated exposures—for example a mortgage and climate change. There are many types of catastrophe risk portfolio concentrated in cities—this granularity can be par- models, from the simplest probabilistic models based upon ticularly important. With this, risk can be underestimated. For historical losses and exposure analysis to more complex and highly location-specific hazards such as flooding, a high level spatially resolved models -used for insurance underwriting - of spatial resolution in both the hazard and exposure data is that use the latest high-resolution climate models to simulate essential to avoid significant biases in risk estimation. large catalogs of realistic events in probabilistic terms and Where such models do not exist, for example, for many de- overlay with detailed exposure data. veloping countries or where is it not feasible to build or obtain such models from proprietary sources, it is common to develop At their core, catastrophe risk models provide a probabilistic risk profiles using public empirical data on natural hazard and view on the financial impacts of weather and climate to par- disaster risk and losses databases (examples include EM- ticular assets, sectors, or economies. They combine science, DAT, DesInventar, and UNEP-GRDP). Various open access engineering, economics, and finance to simulate, in probabi- catastrophe risk models are also now becoming available and listic terms, the potential financial impacts of disasters to a could play an important and growing role in coming years. given portfolio. As noted in Section 1, such tools are used routinely today by insurance supervisors and firms as part of The use of catastrophe risk models for assessing the financial stress testing exercises to ensure the stability and solvency risks of climate change is not new. Such models are already of insurers and are increasingly used by ministries of finance being used to respond to supervisory requirements for climate EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 24 change stress testing of insurance companies in several coun- et al. 2013; Hallegatte et al. 2012). It is also consistent with the tries (PRA 2019). Indeed, so-called climate-conditioned ca- standard requirements for stress testing and vulnerability as- tastrophe risk models have been used by both the insurance sessment by central banks (IMF 2019). As such, it is proposed industry and government to assess economic and financial here as the basis of a framework for scenarios to explore fu- impacts of climate change since circa 2005 (see ABI 2005 for ture physical climate-related financial risks. This is a differ- example). There is a significant amount of experience in the ent approach to that encapsulated in current scenarios that use of climate and catastrophe risk models to develop climate provide deterministic projections of future climate risks based change scenarios (for example, see Golnaraghi 2021). Like all purely on models. An approach that more explicitly recognizes models, catastrophe risk models come with uncertainty (Aerts deep uncertainty is more akin to the Realistic Disaster Sce- et al. 2014), but there is substantial knowledge about how to narios employed by the insurance industry (Lloyd’s of London manage uncertainties in these models in decision making (see 2021; PRA 2019). Dietz and Niehörster 2020). Substantial literature exists on scenario development under In summary, to generate probabilistic scenarios of extreme conditions of deep uncertainty (Lempert et al. 2013). This con- weather impacts toolkits are available that are tried and tested cludes that uncertainty should not be ignored, but instead that within decision making in the financial sector and commonly ap- scenarios be developed to represent the range of possible plied within adaptation decision making and climate risk man- outcomes. Model intercomparison initiatives, like the Climate agement in other sectors. This existing knowledge, particularly Model Intercomparison Project (CMIP11) and the Inter-Sectoral that which has been developed by the insurance industry over Impact Model Intercomparison Project (ISIMIP), attempt to several decades, can be readily deployed to support the de- put bounds on model uncertainty through comparing multiple velopment of acute risk scenarios for banks. For the reasons models run with the same scenarios. However, as noted by described above, probabilistic risk assessment or risk profiles, Fiedler et al. (2021), ranges generated through such exercises including catastrophe risk models, should become an important should not be interpreted as the bounds of future outcomes. part of the arsenal of ministries of finance and central banks to To account for deep uncertainties, scenario generation ex- assess the future financial risks from climate change. ercises will often include model-based projections alongside scenarios developed through expert judgment and the best available science. In the physical climate literature, scenarios 4.2. Fully Capturing the that aim to explore the space of possible future outcomes are referred to as story lines (Jack et al. 2020). Uncertainty in Current Climate and Impact Projections The use of such scenarios is commonplace in climate change adaptation planning; for example, such an approach was adopted in the development of scenarios of sea level rise to A further challenge to be addressed is how to represent the inform the construction of the Thames Barrier that protects uncertainties in climate (and catastrophe risk) models within London from flooding (see review by Ranger, Reeder, and scenario development. With climate change, the frequencies Lowe 2013). In this case, scenarios included not only mean and intensities of weather extremes are expected to shift, of- projected changes in sea level over the coming decades from ten toward more frequent intense events (IPCC 2012). How- climate models, but also a “high+” and “high++” scenario that ever, the scale (and sometimes direction) of changes with cli- represented the potential for low probability but high-impact mate change is deeply uncertain; this means that it is currently outcomes based upon expert judgment and the best available not possible to attach probabilities to such scenarios. science on ice sheet melt. As noted by Schinko et al. (2017) in the context of deep un- Such approaches have been applied to generating scenarios certainty, models and scenarios that allow to “explore rather of extreme weather events with climate change to inform deci- than predict” can better help understand the drivers of indi- sion making (for examples, see Ranger and Niehörster (2012) vidual and system-level responses to shocks in comparison for hurricanes in the Atlantic and Daron et al. 2018 and Gallo with forecasting models. This approach—which aims to de- et al. 2018 for typhoons in the Philippines). Such scenarios velop plausible but severe scenarios to explore vulnerabilities have been used to explore the range of impacts of climate and risk mitigation options—is well developed in other areas change on the insurance industry (Kunreuther, Michel-Kerjan of climate risk management (Kunreuther et al. 2014; Lempert and Ranger 2013). This type of scenario development ap- 11. For more on CMIP see https://www.wcrp-climate.org/wgcm-cmip. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 25 proach appears well aligned with practice in the development Empirical studies and modeling demonstrate that the indirect of scenarios for stress testing. This is similar to the approach impacts can be at least as large as, if not larger than, the di- taken by the International Monetary Fund (IMF) in the Philip- rect impacts of physical climate shocks (Hallegatte, Hourcade, pines (IMF 2021), which built upon those studies noted above. and Dumas 2007; Hallegatte 2019; Colon, Hallegatte, Rent- schler, and Rozenberg 2019; Dunz et al. 2021). The scale of In summary, uncertainties should not be ignored but rather the indirect impacts versus the direct impacts is dependent on a scenarios developed that span the range of plausible out- number of factors, including the level of preparedness and resil- comes. There is an established literature and practice to draw ience of the economy to shocks. Factors include, for example, upon. Scenario generation exercises that combine model- insurance penetration, investments in contingency planning, based projections with expert judgment and the best available and access to labor and credit postdisaster (Koks and Thissen science could also be important in ensuring scenarios fully 2016; Ranger et al. 2011). Botzen, Deschenes, and Saunders represent tail risks. (2019) conclude that while the net macroeconomic (that is, indi- rect) losses are overall negative, they are likely to be small for high-income economies, as they are better able to cope with 4.3. Representing the Indirect negative production shocks and generally more severe for low- income countries and smaller, less-diversified economies. Impacts of Physical Climate-Related Shocks Indirect effects of physical climate shocks can be quantified us- ing computational macroeconomic models. Such models pre- dict the impacts of shocks on a variety of economic indicators, Most existing physical climate risk assessments, including, such as GDP level and growth, trade, and employment. Botzen, for example the NGFS Climate Impact Explorer, provide esti- Deschenes, and Saunders (2019) provide a review of model- mates of the direct impacts of physical climate shocks (Smith ing approaches and find that research on the indirect impacts 2021), but miss the indirect impacts. In many cases these can of natural disasters builds on the predictions of input–output be at least as large or larger. The differences between direct (Hallegatte 2008), computable general equilibrium (CGE) mod- and indirect impacts can be thought of as damage to stock els, and most recently, structural econometric models (Burns, within an economy (capital stock, for example) versus impact Jooste, and Schwerhoff 2021). Integrated Assessment Models on flows, including supply chains and production. (IAMs) have been developed that estimate the impacts of climate change in GDP terms. As described by Botzen, Deschenes, The direct impacts of an event can be defined as the direct cost and Saunders (2019), most IAMs estimate the aggregate eco- of repairing or replacing (at the pre-event price level) assets nomic impacts of climate change, so they do not explicitly rep- that have been damaged or destroyed (Hallegatte 2008).12 resent physical climate shocks. Some IAM applications have, Such costs are routinely estimated by insurance companies however, focused on natural disasters (for example, Narita, Tol, and are output of traditional catastrophe risk models. But im- and Anthoff 2010). More recently, scholars started to recognize portantly, the direct cost of a disaster is often only part of the the need for bottom-up and out-of-equilibrium models rooted on overall economic cost and in some cases, can constitute only complex system science to understand complex and intercon- a fraction of the overall costs to a particular firm and the over- nected sources of systemic risk emerging from the interaction all economy. The remainder is the indirect loss, which was between climate change, the real economy, and the credit and defined by Hallegatte (2008) as “the reduction of total value financial markets (Farmer et al., 2015; Battiston, Farmer, et al. added by the economy because of the disaster; (the indirect 2016). Rezai and Stagl (2016) called for the development of loss is) the reduction in production of goods and services, and a new generation of models in ecological macroeconomics— can include business interruption in the event aftermath, pro- models that are able to integrate the microfoundations of the duction losses during the reconstruction period, and service models with a meso- and macroeconomic level of analysis to losses.” Indirect losses refer to changes in economic activity better understand the feedback loops between the ecosystem, that follow the disaster and include any positive spillover ef- the real economy, and the financial sector. Agent-based models fects due to the substitution of production and the demand and stock-flow consistent models are two families of models for reconstruction (Botzen, Deschenes, and Saunders 2019). that contributed to address these concerns (Monasterolo and This captures both the short- and long-term economic losses Raberto 2018; Monasterolo and Raberto 2019). An illustrative in economic production and consumption and any related eco- application of a stock-flow consistent model, EIRIN, to calculate nomic recovery paths (Kousky 2014). indirect impacts is shown in section 4.4. 12. Botzen, Deschenes, and Saunders 2019 provide a slightly broader definition of direct economic losses to include “the destruction of residences, businesses, productive capital, infrastructure, crops, livestock, and (monetized) physical and mental health impacts” EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 26 Three conclusions emerge relevant to physical climate-related points toward the materiality of these impacts versus the direct financial risk assessment. Firstly, the indirect loss from physi- impacts normally considered within physical, climate-related cal climate shocks are a very material component of finan- risk assessment as well as the substantial additional layer of cial risk, particularly for emerging and developing economies; uncertainty introduced. A challenge is that no one modeling ap- ignoring this could lead to a systematic underestimation of proach captures the full range of indirect impacts. For physi- the risks. Secondly, a set of well-tested tools are available to cal, climate-related, financial risk scenarios, most critical will assess the indirect impacts of physical climate shocks, ap- be to identify those factors most material in terms of financial plicable at local, national, or global scales. Thirdly, this will risks, based on the evidence and knowledge of the context and typically require macroeconomic models that can identify and analyses of transmission channels, and ensure these are ad- quantify the complex transmission channels and feedbacks in- equately reflected in scenario generation and model selection. volved. Importantly, these macroeconomic models should be forced with climate and/or catastrophe risk models that fully As outlined in CFRF 2020, the starting point will be to under- represent uncertainties and tail risks as outlined in the previ- stand key drivers of risk to the area and actors of interest and ous two sections. Using only AALs for example, will lead to to explore the risk transmission channels. From this, it is pos- underestimates. sible to develop scenarios that stress test the key relevant fi- nancial vulnerabilities and to base this upon a combination of However, there are challenges. First, there is not yet a clear macroeconomic models, empirical data, and quantitative and consensus on which models should be used for this type of qualitive scenario-based approaches as appropriate. application, and there is no comprehensive model intercom- parison (similar to a CMIP or ISIMIP) that allows decision mak- ers to assess the uncertainties. A second challenge is that the 4.4. Impacts on the Financial dependence of indirect impacts on the specific characteristics of the economy and the shock mean that it would be difficult to Sector and Economy-Financial draw out some simple relationship, such as an ‘indirect impact Sector Feedbacks vulnerability curve,’ or generate generic scenarios that could be easily applied to any country to avoid the need for tailored, country-specific analyses. To take the final step from a physical climate shock to esti- mate the impact on the financial sector requires an additional It is also important to note that all the macroeconomic models layer of analysis. The available evidence here is limited. One commonly used in climate stress testing do not yet capture all approach taken is to apply empirical relationships between those indirect factors known to be important to physical climate shocks and variables such as nonperforming loan ratios or risk assessment. For example, recently there has been signifi- z-scores to estimate future impacts of climate change on cant work on the economic impacts of infrastructure systems the financial sector (see Klomp 2014). A strong limitation of disruption associated with physical climate shocks that shows such approaches is that current financial risk mitigation mea- that this contributor to indirect economic cost alone can be far sures, such as forbearance, can mask the impacts of shocks larger than the direct impacts on buildings and infrastructure. in historical data. Another approach is to model the impacts Koks et al. (2019) find that 27 percent of all global road and on firms from a reduction in revenues or return on assets as- railway assets are exposed to at least one physical climate sociated with a physical climate shock and the consequent hazard. Hallegatte, Rentschler and Rozenberg (2019) con- increase in debt at risk (Feyen et al. 2017) or probability of clude that altogether, infrastructure disruptions impose costs default (Merton, 1974). However, these approaches do not between $391 billion and $647 billion a year on households take full account of the complex feedbacks between the real and firms in low- and middle-income countries alone. Physical economy and the financial sector. These feedbacks can act climate-related shocks can also send ripple effects through to amplify or dampen the impacts of physical climate shocks global supply chains and patterns of trade, as observed, for on the banking sector (see Section 2); in large higher-income example, during the global food price shock of 2008/10, partly countries with well-diversified financial markets, the impacts of driven by extensive droughts (IEG 2013). historical disasters on the banking sector has been generally limited compared to smaller economies and those with more In summary, there is a large evidence base available on the concentrated financial sectors. The interconnectedness of the indirect impacts of physical climate shocks nationally and glob- real economy and financial sector (for example, firms’ borrow- ally, yet this is not captured in current physical climate scenar- ing and banks’ lending, foreign households’ remittances and ios designed for central banks and supervisors. The evidence domestic households’ disposable income) could contribute EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 27 to reverberate and amplify the original shock in the economy model to illustrate the importance of such feedbacks by turning (Battiston, Caldarelli, et al. 2016; Bardoscia et al. 2021). on and off credit constraints. We utilize the EIRIN model (Mon- asterolo and Raberto 2018; Monasterolo and Raberto 2019), For instance, in the 2008 financial crisis, the shock stemming which is an open economy macrofinancial model composed from mortgages in the US market spread fast to the European of heterogeneous agents and sectors of the real economy and financial sector and the rest of the world due to the exposure finance represented as a network of interconnected balance of European and extra European financial actors to the deriva- sheets. EIRIN is stock-flow consistent (SFC): every agent is tive contracts and institutions hit by the crisis. Then, due to the represented by its balance sheet items, calibrated on real data large role that finance plays in today’s economy, the shock (when possible), making it possible to trace a direct correspon- fast spread in the real economy causing cascading, nonlinear dence between stocks and flows in the economy and finance effects, in particular on private and public debt sustainability. and changes as a result of exogenous shocks (natural disas- COVID-19 showed that a major global systemic shock to the ters, for example) and endogenous shocks (change in policy financial sector can also originate from exogenous factors: and financial regulation, change in investors’ expectations). As in that case, a global pandemic and the policy measures im- a difference from most macroeconomic models used for climate posed to control the health impacts. Here, unlike the global fi- and disaster risk assessment, EIRIN embeds a financial sector, nancial crisis, the shock to the financial sector originated in the financial market, and a central bank in charge of conventional real economy. This has similarities with what one might expect and unconventional monetary policies. In contrast, in traditional from future physical climate shocks (that is, exogenous drivers macroeconomic models—such as real business cycles, com- of macrofinancial risks). Such strong financial feedbacks driv- putable general equilibrium (CGE), and dynamic stochastic en by an exogenous shock were demonstrated by Mandel et general equilibrium (DSGE) models—the role of money and fi- al. (2021) in the context of flood risk. Missing such feedbacks nance is either absent or treated as a friction (Galí 2018; Jakab within scenario development and macrofinancial modeling and Kumhof 2019), thus preventing analysis of endogenous could lead to under- or overestimating financial risks. building up of financial crises and their effects on the economy and policy decisions (Monasterolo 2020). It is also important to note that many countries have financial ‘shock absorber’ mechanisms in place designed to dampen A further advantage of using the EIRIN model in this context the impacts of shocks on firms and the financial sector, includ- is that it is not constrained to solve to equilibrium, thus allow- ing partial credit guarantees schemes or other mechanisms, ing the analysis of the causes and consequences of nonlinear- such as forbearance, adjusting provisioning requirements, in- ity of impacts on economic and financial investments and policy terest rates, or quantitative easing by central banks. It will be decisions. These emerge endogenously from agents’ reaction important to build such mechanisms into scenarios. However, to shocks, considering the interactions among economic and notably such mechanisms often imply risk being taken onto financial agents and sectors. The model allows to account for the government balance sheet that would have implications the richness of risk transmission channels and impacts, consider- for macrofiscal risks as well as imply limits on what risks could ing how the nature of risk affects agents’ heterogeneous beliefs, be mitigated in the future with climate change. intertemporal preferences, formation of expectations, and deci- sion making in response to the shocks. In this case, the EIRIN Capturing the main transmission channels and feedbacks as model is initiated with scenarios with and without a direct physical well as representing any financial shock absorber mecha- climate shock. The direct loss simulation serves as a ‘shock’ to nisms requires a macrofinancial model, that is, a macroeco- the economic system, causing economic interruption and diver- nomic model with sufficient resolution of the financial sector. sion of economic flows, hence a potential loss amplification in Macrofinancial models are now regularly applied to assess the the aftermath of a disaster. In Gourdel et al. (2021), the EIRIN impacts of transition scenarios on the financial sector (Battis- model was advanced to include a stronger representation of the ton et al. 2017; Roncoroni et al. 2021), but examples of their financial sector and market to allow the integration of macrofi- application to physical climate-related financial risks is limited. nancial dynamics into full financial network models (Battiston et Current physical climate-related financial risk assessments al. 2017; Roncoroni et al. 2021) that analyze direct and indirect usually do not incorporate these financial sector feedbacks. losses (such as second, third round impacts) due to financial in- terconnectedness. The transmission channels between the real Given the scarcity of evidence, we provide an illustration of mac- economy, the financial sector, and the public sector are illustrated rofinancial modeling of the impact of a physical climate shock in figure 4.2 and is further described in Gourdel et al. (2021). on a highly exposed middle-income country. We also use this EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 28 > > > F I G U R E 4 . 2 . - Transmission Channels Between the Real Economy, the Financial Sector, and the Public Sector Capital stock, Direct Natural hazard Firms’ Firms’ infrastructure Impact (flood in EU MS) production performance destruction Cascading effects: Cascading effects: Cascading effects: ECONONY PRIVATE FINANCE PUBLIC FINANCE • Investing • Credit Risk (∆PD) • Fiscal revenues • Employment • Cost of capital (interest rate) • Sov. bond spread • Wages • Banks’ financial stability • Interest on debt • GDP (high/low-carbon) (NPL, leverage, LGD) • Sov. devt sustainability • Households’ inequality Feedback Feedback Indirect Impacts Source: Original for this publication. Note: Channels of natural disasters (tropical storm) risk transmission to the economy (blue) and its macroeconomic (light red shadow) impacts on the real economy, financial sector (private finance) and government (public finance) within the modified EIRIN model used in this study. Sov = sovereign. LGD = loss given default. PD = probability of default. GDP = gross domestic product. NPL = nonperforming loans. EU MS = EU Member State. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 29 > > > B OX 4 .1 . Illustrative Scenarios Developed for the EIRIN Model F I G U R E B 4 . 1 . 1 . - Four Scenarios Considered in This Study. Scenario No Natural Hazard Graphical Occurrence Representation Timing: Q4 2020 1 Impact Size: ˤH = 1.63% Strong hazard Agriculture = 0.147% Industry = 0.5058% Time (typhoon) Q2 2020 Q3 2020 Q4 2020 Service = 0.978% 2 COVID-19 No Time emergency Q2 2020 Q3 2020 Q4 2020 3 Timing: Q4 2020 Compound COVID-19 Time Impact Size: ˤL = 0.46% and mild hazard Q2 2020 Q3 2020 Q4 2020 4 Timing: Q4 2020 Compound COVID-19 Time Impact Size: ˤH = 1.63% Q2 2020 Q3 2020 Q4 2020 and strong hazard The figure above illustrates the four scenarios considered in this illustrative study. Scenario 1 (SC1) is characterized by the occurrence of typhoons that hit late in the typhoon season. Scenario 2 (SC2) is characterized by the COVID-19 shock (no typhoon). Scenario 3 (SC3) considers the case of the COVID-19 shock followed by a low-impact (mild) typhoon that occurs late in the typhoon season. Scenario 4 (SC4) considers the case of the COVID-19 shock followed by a high-impact (strong) typhoon that occurs late in the typhoon season. The impact of natural hazard is estimated as relative loss of capital stock by economic sector, based on a fitted Findex damage function relevant to the country, calculated using World Bank in-house catastrophe risk models. Source: Ranger, Mahul, and Monasterolo (2021) and references therein. Figure 4.3, panels a, b, and c illustrate three outputs from the (CAR) represents the ratio between the bank’s equity and EIRIN model relevant to understanding the macrofinancial the banks’ risk-weighted assets (in this analysis, we consid- impacts of physical climate shocks (a strong typhoon repre- er loans) (figure 4.3, panel c). Here, the CAR influences the sented by the orange line on all three graphs) based on Mon- amount that banks can lend to firms, conditioned to the regu- asterolo et al. 2021. The four scenarios employed (SC1, SC2, latory CAR that considers the risk exposure of the bank via SC3, and SC4) are described in Box 4.1. loans. Thus, it represents a maximum credit supply. The CAR falls following the shock, reducing credit supply. This reduction Figure 4.3, panel shows a peak 2.5 percent loss of real GDP peaks at 10 percent below BAU and persists for several quar- occurring in the quarter following the shock and then declin- ters. This analysis demonstrates the materiality of physical ing. This creates an immediate credit shock, including almost climate risks to both economic output and key financial sector a doubling of credit demand from firms for investment and li- soundness indicators, such as the CAR and credit demand. quidity purposes. Credit demand recovers quickly as invest- Explicitly modeling these transmission channels is important ments align with those of the business as usual (BAU) sce- in assessing the scale of financial risks. nario (figure 4.3, panel b). Finally, the capital adequacy ratio EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 30 > > > FIGURE 4.3. Panels a, b, c: Three Outputs from the EIRIN Model Including a Strong Typhoon a: Real GDP Indexed Against the Business as Usual (BAU) Scenario Real GDP index 100.0 97.5 95.0 92.5 90.0 87.5 85.0 82.5 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 2020 2021 2022 2023 2024 Time (Quarters) b: Impacts on Credit Demand by Firms Credit demand 300 250 200 150 100 50 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Time (Quarters) b: Impacts on Credit Demand by Firms Capital Adequacy Ratio (CAR) 102.5 100.0 97.5 95.5 92.5 90.0 87.5 85.0 82.5 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Time (Quarters) BAU SC3 Compound COVID-10 and moderate hazard SC1 Strong hazard (typhoon) SC3 Compound COVID-10 and strong hazard SC2 COVID-19 emergency Source: based on Monasterolo et al. 2021 EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 31 Figure 4.4 illustrates why these feedbacks are important to consider within the development of scenarios to assess physical climate-related financial risks. It shows real GDP for the same shock but under different credit constraints—an important feedback between the real economy and the financial sector. It illustrates that when credit constraints are strong (represented by a high regulatory CAR) the impacts of a physical climate shock on GDP are substantially amplified and more persistent. Such credit constraints could be generated by high demand for credit (particularly the context of compounding shocks illustrated in figure 4.4), changing policies by banks or changes in regulation to protect the financial sector. This illustrates how not representing potential feedbacks between the real economy and the financial sector in scenarios or not allowing these feedbacks in modeling, could cause central banks and other financial institutions to substantially underestimate the potential risks; this undervaluation and pric- ing of risk could have systemic implications. The feedbacks explored here are national only; yet Mandel et al. 2021 also demon- strate the importance of international financial networks as potential amplifiers of risk. > > > F I G U R E 4 . 4 . - Real GDP Indexed Against the BAU Scenario for a Compound Shock (Typhoon plus COVID-19) in the EIRIN Model with Different Credit Constraints Real GDP indexed 100.0 95.0 90.0 85.0 80.0 75.0 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 2020 2021 2022 2023 2024 Time (Quarters) BAU SC4-1 Compound COVID-19 and strong hazard; Regulatory CAR (% of bank CAR) = 75 SC4-2 Compound COVID-19 and strong hazard; Regulatory CAR (% of bank CAR) = 80 SC4-3 Compound COVID-19 and strong hazard; Regulatory CAR (% of bank CAR) = 85 SC4-4 Compound COVID-19 and strong hazard; Regulatory CAR (% of bank CAR) = 90 SC4-5 Compound COVID-19 and strong hazard; Regulatory CAR (% of bank CAR) = 95 Source: based on Monasterolo et al. 2021 EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 32 Battiston et al. (2021) also noted the importance of account- 4.5. Compounding Risks ing for investors’ expectations in the realization of the climate scenarios because they affect the cost of capital and thus firms’ investment decisions. This finance-climate feedback Physical climate shocks will not happen in isolation. Physi- is currently not included by climate scenarios, but it is cru- cal climate shocks will combine with other shocks and stress cial to avoid underestimating risk in stress testing exercises. within the economy, such as economic cycles, pandemics, or Indeed, research has shown that financial actors’ expecta- financial crises. When different types of shocks compound tions and anticipation of climate risks (the so-called climate within an economy, they can generate nonlinear effects that sentiments), affect both the viability and performance of can amplify losses significantly. This is already well recog- investments in high (low) carbon sectors negatively (posi- nized within standard approaches to stress testing, for exam- tively) and thus the success of climate mitigation and adapta- ple IMF 2019 and Lloyd’s of London 2021. For this reason, it tion (Dunz et al. 2021). In this context, investments that are is important to include scenarios that capture compounding considered as crucial for mitigation, could not materialize, risks within physical climate-related financial risk assessment. leading to scenarios that are not considered by the NGFS. In contrast, investments in climate misaligned activities (or In addition, physical climate shocks will interplay with chronic carbon stranded assets) could increase, thus increasing the changes and so cannot be treated entirely independently. For exposure to physical climate-related financial risks. example, sea level rise will increase the risks associated with typhoons and storm surges in coastal regions; strain on global Further work is also needed to explore how financial in- food systems resulting from gradual changes in temperatures stitutions serving more vulnerable groups, such as agrib- and rainfall could increase vulnerabilities to droughts. In addi- anks, MFIs, and SACCOs are affected by physical climate tion, transition and physical climate-related risks will happen in shocks. In particular, it is important to understand through parallel and will combine. which channels physical climate risk becomes material and affects such actors and the potential implications for over- The compounding of shocks of different nature (such as pan- all financial sector development and stability (including con- demics, acute and chronic climate changes, economic crises, tagion effects). In low-income and emerging countries with and financial shocks) represents a new type of risk for mac- an underdeveloped financial sector, the direct implications roeconomic and financial research. When risks compound, for financial stability may be limited: major impacts could be they can generate nonlinear dynamics in the economy and expected on financial services on vulnerable communities. finance, generating a prolonged out-of-equilibrium state of the Indeed, in some economies, these nonbanks represent an economy and potential amplification effects (Monasterolo et important part of the overall financial sector; there may be a al. 2021). The macrofinancial implications of compound risk case of ‘too many to fail.’ This in turn would have important cannot be simply detected by the sum of individual risks so indirect impacts on inequality and on poverty alleviation, with should be explicitly built into scenarios. cascading effects on living conditions of rural communities and their socioeconomic development, which over time could To explore the potential scale of compounding risks in this have systemic effects. context, we applied a second set of scenarios to the EIRIN model that included a physical climate shock, pandemic (CO- The analysis demonstrates that including such direct and in- VID-19) and associated economic shock (linked to locking direct impacts of climate risks shocks is critical to provide a down of economic activities). comprehensive assessment of risk to which specific sectors and segments of society are exposed to and to identify tai- While different combinations of shocks will lead to difficult com- lored policy response via risk mitigation and adaptation. This pound outcomes, we choose the following for illustration, with analysis suggests that scenarios could otherwise underes- referring model simulations shown earlier in figures 4.2 and 4.3. timate the scale of the impacts significantly. The approach proposed above is a first attempt to include such direct and Pandemics and disasters have different direct impacts and indirect economic impacts in a comprehensive climate risk affect respectively demand and supply in the economy. How- assessment. Few models incorporate such feedback, and ever, by impacting simultaneously on the firms’ production and this can be a constraint for advancing climate financial risk household demand, indirect impacts get amplified by agents’ assessment. Of critical importance is, at a minimum, to rec- response to the shocks (and to the potential policy measures ognize this potential gap and explore opportunities to repre- taken in response to the shock), and by agents’ interaction. sent this transmission channel through scenario design. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 33 For example, both COVID-19 and physical climate risk impact on firms’ expectations and investment decisions, yet through dif- ferent channels. This, in turn, can increase unemployment, reduce wages, and reduce household welfare, creating a reinforcing feedback on demand, so amplifying the indirect economic impact. This can lead to long-lasting negative socioeconomic effects on both firms and people and slowed growth and recovery. We can measure this compounding as a compound risk multiplier (figure 4.5) and find that it can peak at over 150 percent in some cases; that is, indirect impacts that can be 50 percent larger than the scale of the sum of the individual shocks. > > > FIGURE 4.5. - Compound Risk Multiplier for Two Example Middle-Income Countries, Where One Is Exposed to a Flood Shock (Country A) and the Other a Typhoon Shock (Country B) During a Pandemic 140 130 120 110 100 90 80 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 2020 2021 2022 Country A Country B Source: Ranger, Mahul, and Monasterolo (2021) The transmission channels and drivers of feedbacks for com- shock scenarios during 2020 due to the direct domestic and pounding shocks are risk specific and can combine in different external shocks in demand and investment. In the scenario ways. As illustrated in figure 4.5 the scale and timing of the characterized by compound COVID-19 and strong natural amplification looks very different between different middle- hazard displays a further spike in credit demand in Q1 2021 income countries depending on the structure of the economy, to rebuild the destroyed capital stock. While demand for credit the timing and nature of the shocks and vulnerabilities to dif- related to the firms’ investment plans increases, investments ferent hazards. In the example above, for both representative can be impaired by supply side constraints in the form of labor country A and B, both large middle-income countries, GDP constraints and credit rationing; this amplifies the impact of the is strongly related to investment and capital stock is working physical climate shock both on investment and GDP. close to capacity, so shocks can have a large indirect impact by damaging capital stock, disrupting economic activity, and This illustration highlights the importance of including sce- reducing investment. Both a disaster and a pandemic impact narios that explore how physical climate shocks compound on production and investment so the compound effect is large. with other shocks and stresses within physical climate-related For country A, the flood shock is more prolonged. financial risk assessments. It demonstrates that this com- pounding effect can significantly amplify the impacts of climate The difference in how the financial sector reacts to these differ- change and so is important for central banks and other finan- ent shocks and the combined effects shown in figures 4.2 and cial institutions to consider within climate stress tests. The 4.3 are also important to note. The compounding shock has a simple approach outlined here could form a model for how much greater and more long-lived impact on GDP. Contrary to compounding risks could be considered within future sets of the physical climate shock, investments drop in the COVID-19 scenarios for central banks. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 34 5. >>> Next Steps Toward Implementation This section suggests an approach to scenario development that aims to take account of the key material risk drivers while remaining practical for real application, particularly in countries where data availability may be constrained. The risk transmission channels are unique to each country and so a tailored approach to scenario development is necessary. Figure 5.1 below illustrates a set of steps to construct climate scenarios for central banks and financial regulators as well as financial institutions and governments. This could take a set of generic scenarios as one input—such as the Network of Central Banks and Supervisors for Greening the Financial System (NGFS) scenarios—combined with a wide range of inputs to ensure that scenarios represent the space of plausible future outcomes tailored to the key risk drivers and transmission channels of a specific country. This would include probabilistic information on extreme weather events and their physical impacts on key sectors and systems, climate change and other future risk scenarios that represent the range of uncertainties in projections. It would also include estimates of indirect impacts for the key risk transmission channels, compound events, and representation of potential real economy to financial sector feedbacks in line with the five risk drivers outlined in Section 4. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 35 > > > F I G U R E 5 . 1 . - A Framework for Scenario Construction for Physical Climate-Related Shocks 1. Diagnose Moterial Risks Analysis of Fiscal and Financial Resilience Based on Historical Analyses + Assessing Current and Historical Weather- Related Risks + Identify Transmission Channels and Exposures to Climate 2. i. Catastrophe Risk Model Probabilistic estimates of physical damages (e.g. damage to capital stock and/or production) from climate-related hazards to a specified Scenario Development or Risk Scenarios Based portfolio of assets/sectors. This should include extremes scenarios, e.g., on Historical Analyses* equibalent to up to at least a 1-in-250 year event. Future climate-related Climate-conditioning and scenario selection scenarios** (from models or other sources) ii. Scenarios of Direct Set of scenarios of physical damages under current and future climate Impacts of Physical conditions. Scenarios are selected to reflect the range of uncertainties Climate Hazards and key risk drivers. Models/scenarios of indirect effects (based on macroeconomic model Set of scenarios of physical (direct + indirect) impacts, including (as or simplified approach) iii. Scenarios of Full relevant) impacts on output. This could include GVA per sector, impats Impacts of Physical on total factor productivity (TFP), impacts on government expenditure, Climate Hazards impacts of macrolevel variables (GDP, employment, savings, and Additional relevant and investment) as is relevant. nonmodeled scenarios from diagnostic Scenarios incorporating potential compounding effects, e.g., with iv. Integrated economic cycles, changing international landscape, other shocks Scenario Set Financial module (pandemics or financial crisis), or ‘background” gradual climate change. (Macrofinancial model or vulnerability curve) Scenarios in financial terms, e.g., impacts on NPLs, capital adequacy v. Financial Risk ratios, credit availabiliy, etc. Scenarios (Could include fiscal aspects: revenues, expenditure, etc). 3. Financial Assessment (varios: stress-testing, financial stability) * It may not be necessary to use a full catastrophe risk model. ** Could in some cases include scenarios of changes to exposure or vulnerability. Source: Original figure for this publication. Note: GDP = gross domestic product; EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 36 To identify and prioritize the key country-specific risk drivers laying financial exposures to openly accessible hazard and transmission channels, the following steps can be maps. followed: • Analyzing the largest possible losses for the economy • Understanding the impacts of historical physical climate and financial actors, considering the characteristics of the shocks and analogous exogeneous shocks, such as pan- financial network such as financial interconnectedness, demics, on the real economy and financial institutions and and including the role of second and third round expo- the key transmission channels, including through empiri- sures (e.g., interbank lending or exposure to the insur- cal analysis. ance industry). • Reviewing a wide range of evidence on the social and Based upon this understanding, it is possible to define a set of economic impacts of physical climate change to the coun- informative, relevant yet pragmatic climate scenarios that span try, including international dimensions of impacts as well the space of plausible future outcomes suitable for climate as vulnerabilities and trends in other key factors that influ- stress testing and scenario analysis. It will also be possible to ence vulnerability to shocks, such as the structure of the identify where the materiality and uncertainties in risks justify economy and urbanization. a greater investment in further analyses, including quantitative • Mapping the exposures of economic and financial activi- modeling such as that presented in this paper. ties from physical climate shocks—for example, by over- EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 37 6. >>> Conclusions This paper identifies potential gaps in current scenarios widely available for physical climate financial risk assessment and aims to improve the understanding and design of physical climate financial risk scenarios, including for climate stress testing. Those gaps represent material financial risks that cannot be ignored within physical climate financial risk assessment. Tools and approaches are readily available and tried and tested to capture these risks within physical climate financial risk assessment. In order to address those gaps, the paper identifies five important areas to consider within a physical climate-related financial risk assessment and scenario design: (1) extreme weather events, (2) uncertainties in climate models, (3) compound scenarios, (4) indirect economic impacts of shocks, and (5) feedbacks between the real economy and the financial sector. The combination of these five areas within a climate-related financial risk assessment using simple scenarios is illustrated through the EIRIN macroeconomic model. The complexity and deep uncertainty of climate change, the heterogeneity of channels of countries’ (financial) exposure to climate risks, and their socioeconomic and financial characteristics, imply new challenges for macroeconomic analysis and stress testing to inform policy making that require further research. However, such uncertainties are not new or unique to climate stress testing and scenario analysis. A ‘Realistic Disaster Scenario’ approach that combines model- based projections, expert judgment and the best available science to develop scenarios relevant to stress testing can help overcome challenges in data scarcity and constraints on availability of models, particularly in emerging markets and developing economies. Such scenarios need not be complex, but instead should aim to represent the material risk drivers and risk transmission channels, the range of plausible outcomes, and their interactions. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 38 Some further challenges can be highlighted. They include: • Strengthening the evidence based on the relationship between the economic impacts of physical climate shocks and the im- pacts on the financial sector. • Quantifying key economic impacts that are currently missing from models, such as the impacts of infrastructure disruptions and the potential for regional and global cascading shocks. • Exploring how physical climate risks interplay with other shocks and stresses, including transition risks, gradual climate change, and feedback with nature and biodiversity as well as economic and financial crises; build this into compound risk scenarios. • Structuring intercomparison exercises between different macroeconomic and macrofinancial models to understand the uncer- tainties in current estimates of economic and financial impacts of physical climate shocks. • Understanding potential contagion effects between larger banks and those serving poor communities that are more heavily affected by physical climate shocks. • Exploring scenarios that take account of long-term trends in exposure, vulnerable and resilient, as well as feedback between climate action and the financial sector over time (Battiston et al. 2021). These are left for further research. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 39 >>> References Aerts, J.C.J.H., W.J.W. Botzen, K. Emanuel, N. Lin, H. de Moel, and E. Michel-Kerjan. 2014. “Evaluating flood resilience strategies for coastal mega-cities.” Science 344: 473–75. Allen, T, S Dees, J Boissinot, G C M Caicedo, V Chouard, L Clerc, A De Gaye, A Devulder, S Diot, N Lisack, F Pegoraro, M Rabaté, R Svartzman and L Vernet. 2020. “Climate-related sce- narios for financial stability assessment: An application to France”, Banque de France, Working Paper Series, no 774, July. Alleyne, T., S. Coleridge, İ. Otker, U. Ramakrishnan, and K. Srinivasan, eds. 2017. Unleashing Growth and Strengthening Resilience in the Caribbean. International Monetary Fund, Washing- ton, DC. Association of British Insurers. 2005. Financial Risks of Climate Change. https://insurance.lbl. gov/documents/abi-climate.pdf. Bardoscia, M., P. Barucca, S. Battiston, F. Caccioli, G. Cimini, G. Garlaschelli, F. Saracco, T. Squartini, and G. Caldarelli. 2021. “The Physics of Financial Networks.” Nature Physics, https:// doi.org/10.1038/s42254-021-00322-5. Battiston, S., D.J. Farmer, A. Flache, D. Garlaschelli, A.G. Haldane, and H. Heesterbeek. 2016. “Complexity Theory and Financial Regulation.” Science, 351 (6275): 818–819. Battiston, S., G. Caldarelli, R.M. May, T. Roukny, and J.E. Stiglitz. 2016. “The price of complex- ity in financial networks.” Proceedings of the National Academy of Sciences, 113 (36): 10031– 10036. Battiston S., A. Mandel, I. Monasterolo, F. Schuetze, and G. Visentin. 2017. “A Climate stress- test of the EU financial system.” Nature Climate Change 7: 283–288. https://doi.org/10.1038/ nclimate3255. Battiston, S., M. Guth, I. Monasterolo, B. Nuerdorfer, and W. Pointner. 2020. “The exposure of Austrian banks to climate-related transition risk.” In: Austrian National Bank (OeNB)’s Financial Stability Report 2020: https://www.oenb.at/en/Publications/Financial-Market/Financial-Stability- Report.html. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 40 Battiston, S., I. Monasterolo, K. Riahi, and B. van Rujiven, 2021. “Accounting for finance is key for climate mitigation pathways.” Science 372 (6545): 918–920. BIS (Bank for International Settlements). 2021. Basel Committee on Banking Supervision. “Cli- mate-related financial risks – measurement methodologies.” BoE (Bank of England). 2020. “Update on the Bank’s approach to the Climate Biennial Explor- atory Scenario in selected areas.” December 16, 2020. Botzen, W.J.W., O. Deschenes, and M. Saunders. 2019. “The Economic Impacts of Natural Di- sasters: A Review of Models and Empirical Studies.” Review of Environmental Economics and Policy 13 (2): 167–188. doi: 10.1093/reep/rez004. Brei, M., P. Mohan, and E. Strobl. 2019. “The impact of natural disasters on the banking sector: Evidence from hurricane strikes in the Caribbean.” The Quarterly Review of Economics and Finance 72: 232–239. Burns, A., C. Jooste, and G. Schwerhoff. 2021. “Macroeconomic Modeling of Managing Hurricane Damage in the Caribbean: The Case of Jamaica.” Policy Research Working Paper No. 9505, Washington, DC: World Bank. https://openknowledge.worldbank.org/handle/10986/34982. Carney, M. 2015. “Breaking the Tragedy of the Horizon–climate change and financial stability.” Speech given at Lloyd’s of London, 29, 220–230. Calice, P. and F. Miguel. 2021. “Climate-Related and Environmental Risks for the Banking Sec- tor in Latin America and the Caribbean: A Preliminary Assessment.” Policy Research Work- ing Paper No. 9694, World Bank: Washington, DC. https://openknowledge.worldbank.org/han- dle/10986/35764. Colon, C., S. Hallegatte, J. Rozenberg, J. 2019. Transportation and Supply Chain Resilience in the United Republic of Tanzania - Assessing the Supply-Chain Impacts of Disaster-Induced Transportation Disruptions. Background study to LIFELINES: The Resilient Infrastructure Op- portunity. Washington, DC: World Bank. Climate Financial Risk Forum. 2020. Climate Financial Risk Forum Guide 2020: Scenario Analy- sis Chapter. June 2020. https://www.fca.org.uk/publication/corporate/climate-financial-risk-fo- rum-guide-2020-scenario-analysis-chapter.pdf Dasgupta, P. 2021. “The Economics of Biodiversity: The Dasgupta Review.” London: HM Trea- sury. Daron, J., I. Macadam, and H. Kanamaru, H., et al. 2018. “Providing future climate projections using multiple models and methods: insights from the Philippines.” Climatic Change 148: 187– 203. https://doi.org/10.1007/s10584-018-2183-5. de Guindos, L. 2021 “Shining a light on climate risks: the ECB’s economy-wide climate stress test.” The ECB (European Central Bank) Blog, March 18, 2021. https://www.ecb.europa.eu/ press/blog/date/2021/html/ecb.blog210318~3bbc68ffc5.en.html. Dietz, S., and F. Niehörster. 2020. “Pricing ambiguity in catastrophe risk insurance.” The Geneva Risk Insurance Review. https://doi.org/10.1057/s10713-020-00051-2. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 41 Dunz, N., A. Essenfelder, A. Mazzocchetti, I. Monasterolo, and M. Raberto. 2021. “Macroeco- nomic and Financial Impacts of Compounding Pandemics and Climate Risks.” Working paper available at SSRN https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3827853. Dunz, N., A. Naqvi, and I. Monasterolo. (2021). Climate sentiments, transition risk, and financial stability in a stock-flow consistent model. Journal of Financial Stability 54: 100872 (in press). ECB (European Central Bank). 2019. “Financial Stability Review, May 2019.” https://www.ecb. europa.eu/pub/financial-stability/fsr/html/ecb.fsr201905~266e856634.en.html. Farmer, J.D., C. Hepburn, P. Mealy, and A. Teytelboym. 2015. “A Third Wave in the Economics of Climate Change.” Environmental and Resource Economics 62: 329–357. Feyen, E., N. Fiess, I. Zuccardi, and L. Lambert. 2017. “Which emerging markets and devel- oping economies face corporate balance sheet vulnerabilities?: a novel monitoring framework (English).” Policy Research Working Paper No. 8198. Washington, DC: World Bank. Feyen, E., R. Utz, I. Zuccardi Huertas, O. Bogdan, and J. Moon. 2020. “Macro-Financial Aspects of Climate Change. Policy Research.” Working Paper No. 9109. Washington, DC: World Bank. https://openknowledge.worldbank.org/handle/10986/33193 License. Fiedler, T., A.J. Pitman, K. Mackenzie, et al. 2021.“Business risk and the emergence of climate analytics.” Nature Climate Change 11: 87–94. https://doi.org/10.1038/s41558-020-00984-6. FSB (Financial Stability Board). 2020. “The implications of climate change for financial stability.” https://www.fsb.org/wp-content/uploads/P231120.pdf. Galí, J., 2018. “The state of New Keynesian economics: a partial assessment”. Journal of Eco- nomic Perspective, 32 (3), 87–112. Gallo, F, J. Daron, I. Macadam, T. Cinco, M. Villafuerte II, E. Buonomo, S. Tucker, D. Hein- Griggs, and R.G. Jones. 2018. “High-resolution regional climate model projections of future tropical cyclone activity in the Philippines.” International Journal of Climatology 39 (3). https:// doi.org/10.1002/joc.5870. Garbarino, N., and B. Guin. 2021. “High water, no marks? Biased lending after extreme weath- er.” Journal of Financial Stability 54: 100874. Golnaraghi, M. 2021. Climate Change Risk Assessment for the Insurance Industry: Research Report. Geneva Association Task Force on Climate Change Risk Assessment for the Insurance Industry. February 25, 2021. Gonzalez Pelaez, A. 2019. Mutual Microinsurance and the Sustainable Development Goals: An impact assessment following Typhoon Haiyan. Cambridge Institute for Sustainability Leadership. Gourdel, R., I. Monasterolo, N. Dunz, A. Mazzocchetti, and L. Parisi. 2021. “Assessing the double materiality of climate risks in the EU economy and banking sector.” Available at SSRN 3939895, forthcoming as European Central Bank working paper. Government of Nepal. 2015. Nepal Earthquake 2015 Post Disaster Needs Assessment. Na- tional Planning Commission. Kathmandu. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 42 Hallegatte, S., J.C. Hourcade, and P. Dumas. 2007. “Why economic dynamics matter in as- sessing climate change damages: illustration on extreme events.” Ecological Economics 62 (2): 330–340. Hallegatte, S. 2008. An Adaptive Regional Input-Output Model and Its Application to the Assess- ment of the Economic Cost of Katrina. Hallegatte, S., A. Shah, R. Lempert, C. Brown, and S. Gill. 2012. “Investment Decision Making Under Deep Uncertainty: Application to Climate Change.” Policy Research Working Paper; No. 6193. Washington DC: World Bank. Hallegatte, S. 2019. “Disasters’ impacts on supply chains.” Nature Sustainability 2, 791–792. https://doi.org/10.1038/s41893-019-0380-5. Hallegatte, S., J. Rentschler, and J. Rozenberg. 2019. Lifelines: The Resilient Infrastruc- ture Opportunity. Washington, DC: World Bankhttps://openknowledge.worldbank.org/han- dle/10986/31805. Hepburn, C. and Farmer, J.D. 2020. ‘Less precision, more truth: uncertainty in climate econom- ics and macroprudential policy’ in Chichilinisky and Rezai (eds) Handbook on the Economics of Climate Change, pp 420-438, 2020 Hosono, K., D. Miyakawa, T. Uchino, M. Hazama, A. Ono, H. Uchida, and I. Uesugi. 2016. “Natural disasters, damage to banks, and firm investment.” International Economic Review 57 (4): 1335–1370. IEG (Independent Evaluation Group). 2013. “The World Bank Group and the Global Food Crisis: An Evaluation of the World Bank Group Response.” Washington, DC: World Bank Group. Inoue, H., and Y. Todo. 2019. “Firm-level propagation of shocks through supply-chain networks.” Nature Sustainability 2: 841–847. https://doi.org/10.1038/s41893-019-0351-x. IMF (International Monetary Fund). 2006. Financial Soundness Indicators Compilation Guide. March 2006. https://www.imf.org/external/pubs/ft/fsi/guide/2006/index.htm. IMF (International Monetary Fund). 2012. “Thailand: Staff Report for the 2012 Article IV Consul- tation.” https://www.elibrary.imf.org/view/journals/002/2012/124/article-A001-en.xml. IMF (International Monetary Fund). 2015 “Nepal: Staff Report for the 2015 Article IV Consulta- tion.” https://www.elibrary.imf.org/view/journals/002/2015/317/002.2015.issue-317-en.xml. IMF (International Monetary Fund). 2019. “The Bahamas: Financial Sector Assessment Pro- gram—Detailed Assessment of Observance of Basel Core Principles for Effective Banking Su- pervision.” IMF (International Monetary Fund) 2021. “Philippines: Financial System Stability Assessment- Press Release and Statement by the Executive Director for the Philippines.” Country Report No. 2021/074. April 9, 2021. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 43 IPCC (Intergovernmental Panel on Climate Change). 2012. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. A Special Report of Working Groups I and II of the Intergovernmental Panel on Climate Change [Field, C.B., V. Barros, T.F. Stocker, D. Qin, D.J. Dokken, K.L. Ebi, M.D. Mastrandrea, K.J. Mach, G.-K. Plattner, S.K. Al- len, M. Tignor, and P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, UK, and New York, NY. 582 pp. https://www.ipcc.ch/site/assets/uploads/2018/03/SREX_Full_Report-1.pdf. IPCC (Intergovernmental Panel on Climate Change). 2014. “Climate Change 2014: Synthesis Report.” Contribution of Working Groups I, II and III to the Fifth Assessment Report of the In- tergovernmental Panel on Climate Change [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland, 151 pp. IPCC (Intergovernmental Panel on Climate Change) 2018. Global Warming of 1.5°C. An IPCC Special Report on the impacts of global warming of 1.5°C above pre-industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development and efforts to eradicate poverty [Mas- son-Delmotte, V., P. Zhai, H.-O. Pörtner, D. Roberts, J. Skea, P.R. Shukla, A. Pirani, W. Moufou- ma-Okia, C. Péan, R. Pidcock, S. Connors, J.B.R. Matthews, Y. Chen, X. Zhou, M.I. Gomis, E. Lonnoy, T. Maycock, M. Tignor, and T. Waterfield (eds.)]. Jack, C.D, R. Jones, L. Burgin, and J. Daron, 2020. “Climate risk narratives: An iterative reflec- tive process for co-producing and integrating climate knowledge.” Climate Risk Management 29: 100239. https://doi.org/10.1016/j.crm.2020.100239. Jakab, Z., Kumhof, M., 2019. Banks are not intermediaries of loanable funds – facts, theory and evidence. Staff Working Paper No. 761, Bank of England. Kalkuhl, M., and Wenz, L. 2020. “The impact of climate conditions on economic production. Evi- dence from a global panel of regions”. Journal of Environmental Economics and Management, 2020, vol. 103, issue C Keys, B.J., and P. Mulder. 2020. “Neglected No More: Housing Markets, Mortgage Lending, and Sea Level Rise.” National Bureau of Economic Research Working Paper 27930. DOI 10.3386/ w27930, October 2020. Klomp, K. 2014. Financial fragility and natural disasters: An empirical analysis. Journal of Finan- cial Stability, 2014, vol. 13, issue C, 180-192 Koks, E. E., and M. Thissen. 2016. “A multiregional impact assessment model for disaster analy- sis.” Economic Systems Research 28: 429–49. Koks, E.E., J. Rozenberg, C. Zorn, et al. 2019. “A global multi-hazard risk analysis of road and railway infrastructure assets.” Nature Communications 10: 2677. https://doi.org/10.1038/ s41467-019-10442-3. Kousky, C. 2014. “Informing climate adaptation: a review of the economic costs of natural disas- ters.” Energy Economics 46: 576–92. Kunreuther, H., E. Michel-Kerjan, and N. Ranger. 2013. “Insuring future climate catastrophes.” Climatic Change 118: 339–354. https://doi.org/10.1007/s10584-012-0625-z. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 44 Kunreuther H., S. Gupta, V. Bosetti, R. Cooke, V. Dutt, M. Ha-Duong, H. Held, J. Llanes-Reg- ueiro, A. Patt, E. Shittu, and E. Weber. 2014. “Integrated Risk and Uncertainty Assessment of Climate Change Response Policies.” In: Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Edenhofer, O., R. Pichs-Madruga, Y. Sokona, E. Farahani, S. Kadner, K. Seyboth, A. Adler, I. Baum, S. Brunner, P. Eickemeier, B. Kriemann, J. Savolainen, S. Schlömer, C. von Stechow, T. Zwickel and J.C. Minx (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY. Kyum Kim, S., and R.B. Peiser. 2020. “The implication of the increase in storm frequency and intensity to coastal housing markets.” Journal of Flood Risk Management 13 (3). https://doi. org/10.1111/jfr3.12626. Lempert, R. J., S.W. Popper, D.G. Groves, N. Kalra, J.R. Fischbach, S.C. Bankes, et al. 2013. Making Good Decisions Without Predictions: Robust Decision Making for Planning Under Deep Uncertainty. Santa Monica, CA: RAND, RB-9701. Lloyd’s of London. 2021. RDS [Realistic Disaster Scenarios] Scenario Specification 2021. Janu- ary 2021. Mandel, A., T. Tiggeloven, D. Lincke, E. Koks, P. Ward, and J. Hinkel. 2021. “Risks on global financial stability induced by climate change: the case of flood risks.” Climatic Change 166 (1): 1–24. McDermott, T., F. Barry, and R. Tol. 2014. “Disasters and development: natural disasters, credit constraints, and economic growth.” Oxford Economic Papers 2014 66 (3): 750¬–773. Merton, R. 1974. “On the pricing of corporate debt: the risk structure of interest rates.” The Jour- nal of Finance 29 (2). https://doi.org/10.1111/j.1540-6261.1974.tb03058.x. Miyakawa, D., and K. Hosono. 2017. “How natural disasters affect banking.” World Economic Fo- rum. https://www.weforum.org/agenda/2014/08/natural-disasters-firm-activity-damage-banks/. Monasterolo, I. 2020. “Embedding finance in the macroeconomics of climate change: research challenges and opportunities ahead.” CESifo Forum 21 (04): 25–32. München: ifo Institut-Leib- niz-Institut für Wirtschaftsforschung an der Universität München. Monasterolo, I., F. Cian, N. Dunz, A. Essenfelder, A. Mazzocchetti, M. Mistry, N. Ranger, and M. Billio. 2021. “Macroeconomic and financial impacts of compounding pandemics, climate and financial risk: a methodological framework.” Working paper. Monasterolo, I., and M. Raberto. 2018. “The EIRIN flow-of-funds behavioural model of green fiscal policies and green sovereign bonds.” Ecological Economics 144: 228-243. https://doi. org/10.1016/j.ecolecon.2017.07.029. Monasterolo, I., and M. Raberto. 2019. “The impact of phasing out fossil fuel subsidies on the low-carbon transition.” Energy Policy 124: 355–370. https://doi.org/10.1016/j.enpol.2018.08.051. Narita, D., R.S.J. Tol, and D. Anthoff. 2010. “Economic costs of extratropical storms under cli- mate change: an application of FUND.” Journal of Environmental Management 53: 371–84. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 45 NGFS (Network of Central Banks and Supervisors for Greening the Financial System). 2020a. NGFS Guide to climate scenario analysis for central banks and supervisors. June 2020. https:// www.ngfs.net/sites/default/files/medias/documents/ngfs_guide_scenario_analysis_final.pdf NGFS (Network of Central Banks and Supervisors for Greening the Financial System). 2020b. NGFS Climate Scenarios for central banks and supervisors. June 2020. https://www.ngfs.net/ sites/default/files/medias/documents/820184_ngfs_scenarios_final_version_v6.pdf NGFS (Network of Central Banks and Supervisors for Greening the Financial System). 2021a. Progress report on bridging data gaps. June 2021. NGFS (Network of Central Banks and Supervisors for Greening the Financial System). 2021b. NGFS Climate Scenarios for central banks and supervisors. June 7, 2021. https://www.ngfs.net/ en/ngfs-climate-scenarios-central-banks-and-supervisors-june-2021. Noth, F., and U. Schüwer. 2017. “Natural Disasters and Bank Stability: Evidence from the U.S. Financial System.” SAFE (Sustainable Architecture for Finance in Europe) Working Paper, No. 167. http://dx.doi.org/10.2139/ssrn.2921000. Perez, J.B. 2015. “Davao Region’s economy records the fastest growth in 2014.” Philippine Sta- tistics Authority. July 30, 2015. Accessed at https://psa.gov.ph/grdp/grdp-id/52664. PRA (Prudential Regulatory Authority). 2019. General Insurance Stress Test 2019. Scenario Specification, Guidelines and Instructions. June 18, 2019. Bank of England. Ramcharran, H. 2017. “Financing small and medium-sized enterprises in Thailand: The impor- tance of bank loans and financing diversification.” The Journal of Entrepreneurial Finance, Vol- ume 19, Issue. 2. The Academy of Entrepreneurial Finance, Montrose, CA. Ranger, N., S. Hallegatte, S. Bhattacharya, et al. 2011. “An assessment of the potential im- pact of climate change on flood risk in Mumbai.” Climatic Change 104: 139–167. https://doi. org/10.1007/s10584-010-9979-2. Ranger, N., and F. Niehörster. 2012. “Deep uncertainty in long-term hurricane risk: Scenario generation and implications for future climate experiments.” Global Environmental Change 22 (3): 703–712. Ranger, N., T. Reeder, and J. Lowe. 2013. “Addressing ‘deep’ uncertainty over long-term climate in major infrastructure projects: four innovations of the Thames Estuary 2100 Project.” EURO Journal on Decision Processes 1, (3–4): 233–262. https://doi.org/10.1007/s40070-013-0014-5. Ranger, N., O. Mahul, and I. Monasterolo. 2021. “Managing the financial risks of climate change and pandemics: What we know (and don’t know).” One Earth 4 (10): 1375–1385. Rezai, A., and S. Stagl. 2016. “Ecological macroeconomics: introduction and review.” Ecological Economics 121: 181–185. Rismanchi, K., J. McDaniels, and S. Gringel. 2021. Prudential Pathways: Industry Perspectives on Supervisory and Regulatory Approaches to Climate related and Environmental Risks. Insti- tute of International Finance. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 46 Roncoroni, A., S. Battiston, L.O. Escobar-Farfán, and S. Martinez-Jaramillo. 2021. “Climate risk and financial stability in the network of banks and investment funds.” Journal of Financial Stabil- ity 54: 100870. Schinko, T., G. Bachner, S.P. Schleicher, and K.W. Steininger. 2017. “Modeling for insights not numbers: The long-term low-carbon transformation. “Atmosfera 30: 137–161. https://doi. org/10.20937/ATM.2017.30.02.05. Smith, P. 2021. “The Climate Risk Landscape: A comprehensive overview of climate risk assess- ment methodologies.” UNEPFI (United Nations Environment Programme Financial Initiative). February 2021. Stern, N. 2006. The Economics of Climate Change: The Stern Review. Cambridge University Press. Stern, N. 2016. “Economics: Current climate models are grossly misleading.” Nature 530: 407– 409. doi:10.1038/530407a. Strobl, E. 2009. “The Macroeconomic Impact of Natural Disasters in Developing Countries: Evi- dence from Hurricane Strikes in the Central American and Caribbean Region.” Proceedings of the German Development Economics Conference, Frankfurt A.M. 2009, No. 35, Verein für Socialpolitik, Ausschuss für Entwicklungsländer, Göttingen. TCG (Thai Credit Guarantee Corporation). 2016. TCG 25th Anniversary, Continue to Driv- ing SME’s Success—Annual Report 2016. Accessed at https://www.tcg.or.th/uploads/ file/200323160301AxCK.pdf UNEP (United Nations Environment Programme). 2021. Emissions Gap Report 2021. https:// www.unep.org/resources/emissions-gap-report-2021 UNEPFI (United Nations Environment Programme Financial Initiative). 2020. “Charting a New Climate.” TCFD Banking Programme Report. September 2020. Vermeulen, R, E Schets, M Lohuis, B Kölbl, D J Jansen and W Heeringa. 2019. “The Heat is on: a framework for measuring financial stress under disruptive energy transition scenarios”, DNB Working Paper, no 625, February. Watkiss, P., T. Downing, et al. 2005. The Social Costs of Carbon (SCC) Review—Methodologi- cal Approaches for Using SCC Estimates in Policy Assessment, Final Report. AEA Technology Environment. Weyant, J. 2017. “Some contributions of integrated assessment models of global climate change.” Review of Environmental Economics and Policy 11 (1): 115–137. World Bank. 2012. “Thai Flood 2011: Rapid Assessment for Resilient Recovery and Reconstruc- tion Planning.” World Bank, Bangkok. World Bank Group. 2014. “Reconstruction and Recovery Planning in the Aftermath of Typhoon Haiyan (Yolanda).” Washington, DC: World Bank. https://openknowledge.worldbank.org/han- dle/10986/21520. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 47 World Bank Group. 2019. Boosting Financial Resilience to Disaster Shocks : Good Practices and New Frontiers. World Bank Technical Contribution to the 2019 G20 Finance Ministers’ and Central Bank Governors’ Meeting;. World Bank, Washington, DC. World Bank. https://open- knowledge.worldbank.org/handle/10986/31887 License: CC BY 3.0 IGO World Economic Forum. 2021. The Global Risks Report 2021. jttps://www.weforum.org/reports/ the-global-risks-report-2021. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 48