Methodological Principles of Insurance Stress Testing – Climate Change Component PDF

METHODOLOGICAL PRINCIPLES OF INSURANCE STRESS TESTING – CLIMATE CHANGE COMPONENT EIOPA-BOS-21/579 27 January 2022 Abbreviations BCBS Basel Committee for Banking Supervision BE best estimate BS balance sheet CIC complementary identification code CQ...

METHODOLOGICAL PRINCIPLES OF INSURANCE STRESS TESTING – CLIMATE CHANGE COMPONENT EIOPA-BOS-21/579 27 January 2022 Abbreviations BCBS Basel Committee for Banking Supervision BE best estimate BS balance sheet CIC complementary identification code CQS credit quality step D&A deduction and aggregation DTA deferred tax asset DTL deferred tax liability EBA European Banking Authority EIOPA European Insurance and Occupational Pensions Authority ESRB European Systemic Risk Board GDP gross domestic product GWP gross written premium IAIS International Association of Insurance Supervisors ICP (IAIS) insurance core principles EU European Union LTG long-term guarantees NACE nomenclature of economic activities Nat-Cat natural catastrophe NCA national competent authority NGFS Network of Central Banks and Supervisors for Greening the Financial System OF own funds ORSA own risk and solvency assessment RFR risk-free rate RM risk margin SII Solvency II SCR solvency capital requirement ST stress test TA total assets TP technical provisions UL/IL unit-linked and index-linked 1 Introduction TABLE 1-1 KEY ASSUMPTIONS AND UNCERTAINTIES SURROUNDING CLIMATE CHANGE SCENARIOS Key assumptions and Macroeconomic Macroeconomic Financial stability Financial stability uncertainties physical transition physical transition Determines the speed and timing of transition, Determine the extent of Determine the speed Determine the extent of Future climate policy and also may have warming and timing of transition warming diffuse impacts on different sectors (for 1 This paper is part of EIOPA’s broader sustainability agenda to integrate environmental, social and governance (ESG) risk assessment in the regulatory and supervisory framework. EIOPA is committed to supporting the European insurance and occupational pension sectors in their transition to climate neutrality and to deliver on the ‘Green Deal’ initiated by the European Commission. 2 Taking into account that similar work are currently carried out in various international fora (IAIS, NGFS, GFIA, etc.), this paper has a provisional nature and there is a possibility that it will be updated in the coming years in the light of new developments given the current discussions taking place worldwide. 3 Climate change has long term impacts but it has also impacts already today. For example, the recent extreme events in 2021 can be linked to climate change. 4 The structural, non-linear and irreversible impact of climate change in the long run has also been referred to as the Tragedy of the Horizons (Mark Carney, Breaking the Tragedy of the Horizon – Climate Change and Financial Stability, 2015): while the physical impacts of climate change will be felt over a long-term horizon, the time horizon in which financial, economic and political players plan and act is much shorter.  example, a widespread carbon tax) Key technologies (for example carbon capture and storage) will be Rate of progress in Could reduce costs or Determine the extent of Determine the extent of particularly important carbon-neutral actually result in an warming warming for some sectors, and technology increase in GDP result in less disruption to existing business models Key assumptions (e.g. Economy may be Financial stability risks Financial stability risks Feedback loops within about GDP) are often affected indirectly could be exacerbated could be exacerbated the model taken as external in the through second-round by second-round by second-round model effects impacts impacts Higher level of adaption Higher level of adaption More diversified More diversified could lower the long- could lower the long- economies, adaptive economies, adaptive Level of adaption and term physical damages term physical damages firms, and resilient firms, and resilient adaptive capacity but might entail higher but might entail higher financial systems could financial systems could adaption costs in the adaption costs in the reduce transition costs reduce transition costs short-term short-term Damages may be higher Damages may be higher than expected, either than expected, either Higher-than-expected Higher-than-expected Non-linear impacts / through direct losses to through direct losses to damages could impacts damages could impacts uncertainties in climate particular sectors or particular sectors or the speed and timing of the speed and timing of modelling through general through general climate policy climate policy macroeconomic macroeconomic channels channels Source: NGFS (2019) 1.1 Climate change risk and transmission channels TABLE 1-2 OVERVIEW OF MAIN TRANSMISSION CHANNELS FOR CLIMATE CHANGE-RELATED RISKS Type of Transmission channel Balance sheet Example Covered in risk impact this paper? Underwriting risk Liabilities Higher than expected insurance claims on damaged insured assets (non-life) or higher than Yes expected mortality or morbidity rates (life/health) Market risk Assets Impairing of asset values due to financial losses affecting profitability of firms, due to for instance Yes business interruptions, or damage to real estate. Physical risk Specific example: equity price shocks Credit risk Assets Deteriorating creditworthiness of borrowers/bonds/counterparties/reinsurers due Yes to financial losses stemming from climate change Specific example: bond price/yield shock Operational risk Assets Disruption of own insurance activities and/or No assets, such as damage to own property Liquidity risk11 Assets / Unexpected higher payouts and/or lapses as Liabilities broader economic environment deteriorates No (not as part of climate ST) Market risk Assets Impairment of financial asset values due to low- Transition carbon transition, for instance stranded assets, Yes ‘brown’ real estate and/or decrease in value of carbon/GHG intensive sectors. Specific example: equity price shock  Credit risk Assets Deteriorating creditworthiness of borrowers/bonds/counterparties as entities that Yes fail to properly address transition risk may suffer losses Specific example: bond price/yield shock Underwriting risk Liabilities Decrease of underwriting business due to No increase of insurance prices in response to higher than expected insurance claims (non-life) or changes in policyholders’ expectations and behavior related to sustainability factors (e.g. green reputation) (life) Underwriting risk Liabilities Higher than expected claims on professional indemnity cover, as parties are held accountable No Legal liability risk for losses related to environmental damages caused by their activities Legal/reputational Assets / Insurers could be held responsible for climate risk Liabilities change and/or not doing enough to No mitigate/adapt 1.2 Elements of a Climate Change Stress Test exercise FIGURE 1-1 STYLIZED OVERVIEW OF CLIMATE CHANGE STRESS TEST ELEMENTS Define specific ST objective Objective and Choose appropriate scope/participants scope (e.g. solo/group, life/non-life, transition/physical/liability risk) Scenario design Define specific climate scenarios narratives and narrative Consider appropriate time horizon and granularity Develop scenario specifications Derive climate Derive impact on climate and financial and financial variables stemming from climate risk (shocks variables on assets and liabilities) Define application of shocks Evaluate financial andrelevant evaluation metrics impact Participants calculate impact on assets and liabilities Forward looking Assess resilience assessment to and potential evaluate implications responses for business models and insurability of risk 2 Objective of Climate Change stress test TABLE 2-1 OVERVIEW OF POSSIBLE OBJECTIVES FOR A CLIMATE ST Microprudential objectives Macroprudential objectives Assess vulnerabilities and resilience of overall Assess vulnerabilities and resilience of individual (re)insurance sector and potential systemic climate (re)insurers to climate change risks and assess size of change risks potential financial exposures/losses to adverse Assess potential spill-overs to other financial sectors climate scenarios and the real economy of climate change risks Enhance understanding of potentially long-term Assess potential implications for future insurability of climate change risks and implications for business risks and potential protection gap for the real models economy related to climate change risks/perils Enhance risk management capabilities to assess and mitigate climate change risks 3 Scenario design 3.1 General principles and scenario narratives Principle 1: given their distinct but interlinked nature, both transition risk and physical risk should ideally be assessed in conjunction in a climate change stress test; Principle 2: given the wide range of possible future climate paths, it is important to consider a range of climate change scenarios and transition pathways that capture different combinations of physical and transition risk. Applying multiple scenarios also allows to take into account different key dimensions, such as the role of climate policy; Principle 3: ST scenarios should focus both on a central path climate projection and on adverse tail events, to assess whether the financial system and insurers are resilient in case of disruptive climate and transition scenarios; Principle 4: scenarios should entail information (ideally quantitative) about climate pathways (key changes in climate factors) and associated financial impacts at a sufficiently granular level. The scenarios should also allow for the identification of key variables/assumptions that affect scenario pathways; Principle 5: scenarios should cover appropriate time horizons to assess the long-term impact of climate change related risks, given the more long-term nature of climate scenarios, while allowing flexibility to derive short-term stress periods from long-term scenarios. The total level of mitigation of climate change risks or, in other words, how much action is taken to achieve Paris agreement goals and reduce greenhouse gas emissions (leading to a particular climate outcome); Whether the transition occurs in an orderly or disorderly way, i.e. are the actions sudden and unanticipated. FIGURE 3-1 STYLIZED CLIMATE SCENARIOS WITH TRANSITION AND PHYSICAL RISKS Source: NGFS Comprehensive report “A call for action: Climate change as a source of financial risk. Early policy action, orderly transition scenario where the transition to a carbon‑neutral economy starts early and the increase in global temperature stays below 2⁰C, in line with the Paris Agreement. Physical and transition risks are minimized in this scenario; Late policy action, disorderly transition scenario where the global climate goal is met but the transition is delayed and must be more severe to compensate for the late start. In this scenario, physical risks arise more quickly early on and transition risks are particularly pronounced compared to the early policy action scenario; Too little, too late scenario, where the manifestation of physical risks spurs disorderly transition, but not enough to meet Paris agreement goals. Physical and transition risks are both high and severe; Business as usual, no additional policy action scenario (‘Hot house world’) where no policy action which has already been announced is delivered. Therefore, the transition is insufficient for the world to meet the Paris agreement climate goal and physical risks will be particularly pronounced. FIGURE 3-2 STYLIZED PATHWAYS FOR POSSIBLE CLIMATE SCENARIO NARRATIVES Source: Bank of England (2019): The 2021 biennial exploratory scenario on the financial risks from climate change 3.2 Scenario specification and granularity of technical specifications FIGURE 3-3 GRANULARITY OF SCENARIO SPECIFICATION Source: EIOPA adapted from Bank of England. TABLE 3-1 ADVANTAGES AND DISADVANTAGES OF DIFFERENT SCENARIO GRANULARITY FOR BOTTOM-UP STRESS TESTING Aggregation level Advantages Disadvantages Simplicity: requires less detail in the specifications Greater flexibility reduces modelling consistency and can be clearly linked to climate research and comparability across firms Allows flexibility for firms to use different models More difficult for participants to calculate impact on Scenario narrative Forces firms to enhance modelling/risk financial metrics management capacity to assess impact of high-level Results can be difficult to validate climate scenarios Only climate variables would have to be specified, Greater flexibility reduces modelling consistency which can be clearly linked to climate research and comparability across firms Allows flexibility for firms to use different models, More difficult for participants to calculate the but achieves more consistency concerning the impact on financial metrics Climate factors impact on key climate factors Results can be difficult to validate Forces firms to enhance modelling/risk management capacity in order to translate climate factors into financial impacts  Ensures consistency not only on climate factors, but Firms would still have to model implications from also on the macroeconomic impact and key broad economic factors to their specific portfolio economic variables (reducing consistency/ comparability) Broad economic Macroeconomic models can be used to estimate Uncertainty regarding model calibration factors broad economic impacts Broad economic factors do not distinguish between economic sectors, which could be impacted quite differently Provides clarity on the implications for different No commonly accepted methodology yet to economic sectors and takes into account different estimate sectoral impacts of climate scenarios impacts across economic sectors (challenging to bridge climate models to economic Classifications are readily available (for instance sector impact) Sectoral NACE 2, GICS or GLEIF) Sectoral impacts do not take into account firm’s Results can be compared against similar studies heterogeneity within sectors Requires mapping of the portfolio to economic sectors Takes into account firm-heterogeneity and specifies Very complex specification and requires extensive firm-specific impacts based on underlying activities mapping of the portfolio to individual assets based on activity calculate impact Ensures comparability /consistency as impacts are Relevant climate data at individual firm level data is provided at individual asset level often incomplete and only provides a partial view on Firm Promotes risk awareness at counterparty level consolidated firm activities Less incentives for capacity/risk management building for firms to assess exposures of individual assets/counterparties, as impacts would be provided to them at a very granular level Specifies impacts at the most granular level Requires highly granular information on underlying Incentives firms to assess climate exposures of economic activities of firms and how these activities assets based on the underlying activity would be impacted by climate change Activity Data on underlying activities is often not available and only provides a partial view on consolidated firm activities Sectoral level for corporate bonds, equities and real estate exposures. For specific sectors a higher granularity may be explored if needed (for instance based on technology used in energy production, e.g. coal, gas, oil or renewables); Country level for government bonds exposures; Regional level for climate related factors, such as temperature and emission pathways and intra-country regional level for climate-related perils. 3.3 Time horizon and treatment of balance sheets TABLE 3-2 OVERVIEW OF CLIMATE CHANGE RELATED RISKS AND EXPECTED TIMING OF EFFECTS Type of risk Timing of effects Financial impact Extreme climate Short to medium term Unanticipated shocks to physical Physical risk events assets, economic distress, possible systemic disruption Gradual Medium to long term Anticipated shocks to physical and warming20 financial assets Anticipated shocks to financial and non-financial (e.g. long-term impacts on profitability of climate sensitive sectors) Short to medium term Unanticipated shocks to financial Transition risk assets and potential stranded assets Source: Adapted from NGFS Technical Supplement to First Comprehensive Report (2019) The frequency of the calculation (i.e. whether calculations are required at intermittent intervals within the modelling horizon); Static/Fixed reference BS without reactive management actions or dynamic BS with reactive management actions (instantaneous shocks to reference BS versus dynamic BS)21. TABLE 3-3 POSSIBLE APPROACHES FOR THE FIXED/DYNAMIC BALANCE SHEET Fixed/ Frequency of the Dynamic balance Outcome Pros Cons calculation sheet At end of Fixed, impact on Climate scenario Relatively easy to Reactive management modelling horizon reference date modelled over short, implement actions/responses not only balance sheet medium, or long term Enhanced considered which could with instantaneous comparability overstate the impact shocks to balance sheet Allows to assess the at reference date, no potential impact reactive management given current actions allowed business/balance sheets Dynamic, balance Climate scenario Reactive Reduces comparability, sheet allowed to modelled over short, management as reactive management change medium, or long term actions/responses actions can vary and may with instantaneous taken into account, be hard to validate shocks to balance sheet more realistic, 19 The EIOPA Discussion Paper on Methodological Principles for Insurance Stress Testing distinguishes between embedded management actions and reactive management actions (Box 2.1 in the respective paper). In the context of climate change, the focus is on reactive management actions: actions that would be taken by undertakings in direct response to a climate change scenario and that are not assumed to be applied in the baseline scenario. 20 One drawback of using gradual warming is the potential non-linear impact on climate change extremes for example. 21 To ensure the plausibility and the consistency of the enforced reactive management actions against the designed adverse scenarios limitations in the applicable reactive management actions might be prescribed.  at with reactive notably for long-term Impact of reactive management actions impacts management actions allowed Allows to assess difficult to assess impact of reactive depending on time management horizon actions/responses At intermittent Fixed, impact on Climate scenario Medium complexity Reactive management intervals (for reference date modelled over short, Allows assessing actions/responses not instance 1 year or 5 balance sheet medium, or long term impacts on current considered which could year intervals) with instantaneous balance sheet over overstate the impact shocks to balance sheet time Adds additional scenario at reference date for specification and specific intervals, no computational burden reactive management compared to only end-of actions allowed period impact Dynamic, balance Climate scenario Reactive Highly complex both in sheet allowed to modelled over short, management actions terms of scenario change medium, or long term and responses taken specification and with shocks to balance into at each interval, computational burden, sheet at reference date more realistic full blown multi-period for specific intervals, Allows to assess ST with reactive reactive management Reduced comparability as management actions actions and responses results will be very hard allowed at each interval to validate (e.g. shock T=10 compared to balance sheet at T=5) a medium-to-long term horizon (e.g. 15 to 30 years); shocks modelled as instantaneous to the reference date BS; a twofold exercise based on fix and a dynamic / constrained balance sheet; collection of qualitative information on the evolution of climate change impact on the business models of insurers; to be assessed at the end of the modelling horizon. Intermediate positions (e.g. in the middle of the time horizon) might be considered based on cost-benefit analysis. 22 The 2020-2021 Banque de France / ACPR Climate Exercise that had a in the 30-years’ time horizon and dynamic balance sheet represent a potential evolution of the current proposal. 3.4 Way forward Multiple climate scenarios to be evaluated focusing on different climate outcomes/scenario narratives, given the uncertainty of future climate outcomes and to allow a range of different combinations of physical and transition risks. While this would add operational and computational burden to the ST exercise (as participants would have to calculate the impact of multiple, distinct climate scenarios), using multiple scenarios allows to take into account different key dimensions of climate change risks and better assess vulnerabilities and resilience to adverse climate scenarios. Scenario and technical specifications with specific climate variables at regional (intra-country) level for perils and financial impacts at a sectoral level (for corporate bonds, equities and real estate)23 and country level (for government bonds), to ensure a balance between complexity and comparability. Methodologies for deriving, specifying and calibrating these variables will be discussed in more detail in sections 4.1 (for transition risk) and 4.2 (for physical risks). A more granular scenario specification, for instance at individual asset/firm level, would be seen as too complex and burdensome at this stage for a bottom-up ST exercise, but will be considered further as part of EIOPA’s work on top-down methodologies and sensitivityanalysis on climate risks. A medium-to-long-term time horizon, with end-of-modelling horizon scenario impact evaluated as an instantaneous shock to the reference BS under a fixed and constrained framework. This allows assessing the potential long-term financial impact of climate change related risks given current business models and BSs. As such it can give an important indication of the size of potential exposures, and hence the required transformation given current business models, should a specific climate scenario materialize, given the more long-term nature of climate scenarios. A separate forward-looking assessment designed to capture the reactive management actions/responses to climate change-related risks to identify the risk mitigation responses that are considered by insurers in response to climate change and better understand the implications of these responses on insurers’ business models, their resilience and the potential spill-over effects (see section 6). This may be enhanced with questions designed to solicit information on the level of integration of climate change-related risks in areas such as governance, strategy, risk management and metrics and targets of insurers. 4 Modelling approaches 4.2.3 Way forward 4.3 Specification and Application of shocks TABLE 4-6 OVERVIEW OF KEY VARIABLES TO BE SPECIFIED IN CLIMATE ST SCENARIO Climate variables Financial variables Physical risk Transition risk Macroeconomic Financial markets Global and regional Emission pathways GDP (aggregate and Government bond yields temperature pathways (aggregate and disaggregated by Corporate bond yields, disaggregate across economic sector and Frequency, severity and disaggregated by world regions and country) correlation of specific and economic sector economic sectors) material climate-related Interest rates (RFR)48 Equity indices/shocks, perils for different Carbon price pathways regions (for non-life) Inflation disaggregated by Commodity and energy economic sector Mortality / morbidity prices, by energy source Residential and parameters (for life) commercial real estate Energy mix prices Treatment of reinsurance Impact calculated gross of reinsurance (i.e. reinsurance treaties are not taken into account for the calculation of the financial impact); Impact calculated both gross of reinsurance and net of reinsurance; Impact calculated net of reinsurance, but with shock to reinsurance recoverable; Impact calculated net of reinsurance. 4.3.1 Way Forward – for the transition risk in terms of shocks to prices / yields to the specific asset classes – for the physical risks, on the liability side in terms of change in the best estimate assumptions or discontinuance of parameters used in the estimation of the technical provision, on the assets side in term of change in value of asset classes pending on the development of a robust methodology. 5 Metrics for evaluation 5.1 Balance sheet indicators TABLE 5-1 BALANCE SHEET INDICATORS BY TYPE OF RISK Indicator Type of risks Notes Excess of Asset over Liabilities (change of) Physical and transition Asset over Liabilities (change of) Physical and transition Stressed value or price change for each of Only transition Only for assets mapped to climate relevant the identified assets (or class of assets) or sectors, physical assets and their related change in portfolio market evaluation technologies. Relative change of total technical Only physical Only non-life business could be considered provisions unless the scenario include also the impact of a change in mortality/morbidity 5.2 Profitability indicators TABLE 5-2 PROFITABILITY INDICATORS BY TYPE OF RISK Type Indicator Type of risks Notes Main Loss Ratio Only physical Overall or split by relevant lines of business Ancillary Overall impact on the firm’s Physical and profit and loss transition Ancillary Impact on the firm’s technical Only physical (for Overall or split by relevant lines result non-life insurers); of business both (for life insurers) 5.3 Technical indicators expected losses – typically average annual losses (AAL) or median losses to show how average losses might change due to the impact of climate change; tail losses – showing how the losses that might be expected in an extreme year could move as a result of climate change. TABLE 5-3 TECHNICAL INDICATORS BY TYPES OF RISKS Type Indicator Type of risks Notes Gross/ceded/net Only physical Main aggregated losses Exposures (Sum Assured) Only phiyical Overall (baseline figures). As ancillary Main information and only if available, split by event49/geographical area Total assets subject to Only transition Baseline figures. Overall or split by sector Main transitional risks or technology Probable maximum loss Only physical It shows the value of the largest loss that Main (PML) is considered likely to result from an event Annual Probability of Only physical It shows the probability that, over a Main occurrence period of one year, an event of a given magnitude occurs. 1 in X years AEP (aggregate Only physical It shows the maximum amount of losses Only for IM users exceedance probability) caused by all the events over a period of one year, corresponding to the given probability level Annual Average Loss (AAL) Only physical It shows the average losses from Only for IM users property damage experienced by a portfolio per year50. 1 in X years Return period Only physical It shows the magnitude of an extreme Only for IM users event (for instance an event with a 1-in- 100 year return period has a 1% chance of being exceeded by a higher magnitude event in any year) Return period of gross Only physical Ancillary losses 5.3.1 Way forward 6 Second-round effects, spillover and forward looking assessment 38/57