Climate scenario analysis for pension schemes: a UK case study

Abstract

This paper demonstrates how climate scenario analysis can be used for forward-looking assessment of the risks and opportunities for financial institutions, using a case study for a UK defined benefit pension scheme. It uses a top-down modelling tool developed by Ortec Finance in partnership with Cambridge Econometrics to explore the possible impacts of three plausible (not extreme) climate pathways of the scheme’s assets and liabilities. It finds that the funding risks are greater under all three climate pathways than under the climate-uninformed base scenario. In the absence of changes to the investment strategy or recovery plan, the time taken to reach full funding is increased by three to nine years. Given that most models currently used by actuaries do not make explicit adjustments for climate change, these modelled results suggest it is quite likely that pension schemes are systematically underestimating the funding risks they face.

Disclaimer

Other than the material expressly identified in this paper as being the intellectual property and copyright of Ortec Finance, the IFoA owns any and all intellectual property rights, including copyright in this paper and its content is protected by copyright.

You are permitted to use and/or view this paper provided that (i) you do not modify the content in any way; (ii) you do not use this paper or any part(s) of it in a misleading context; (iii) your use of this paper is for your own personal information or for a non-commercial purpose; (iv) any copies of this paper or any part(s) of it for a permitted purpose described in this notice will include an acknowledgement that copyright belongs to the IFoA and Ortec Finance (as appropriate).

The IFoA and Ortec Finance do not accept any responsibility or liability to any person for loss or damage suffered as a consequence of their placing reliance upon any view, claim or representation made in this paper. The information, including the modelling scenarios, and expressions of opinion contained in this paper are not intended to be a comprehensive study, nor do they provide actuarial advice or advice of any nature and should not be treated as a substitute for specific advice concerning individual situations. We have used this tool to model the effects of three different climate pathways through sets of alternative economic assumptions that mimic various possible evolutions of the economy depending on which climate pathway the world follows.

Executive Summary

The potential influence and risk of climate change on economies and our financial system is no longer in doubt. Climate-informed decision-making is increasingly being demanded by regulators. It is rapidly becoming the case that trustees of UK defined benefit pension schemes will be deemed to be not acting responsibly if they do not take any action.

As part of this, there has been an explosion of interest in climate scenario analysis due to its prominence in the TCFD (Taskforce on Climate-related Financial Disclosures) recommendations. Furthermore, amendments to the Pension Schemes Bill currently before Parliament are expected to make TCFD style reporting – and hence climate scenario analysis – mandatory for large UK pension schemes.

This paper explores how climate scenario analysis can be used for forward-looking assessment of the risks and opportunities for defined benefit pension schemes and other financial institutions. For this paper, we have used ClimateMAPS, the top-down modelling tool developed by Ortec Finance in partnership with Cambridge Econometrics. We have then translated the impacts of climate-adjusted GDP from this tool for the following three different climate pathways onto a wide range of financial and economic variables:

• Paris Orderly (coordinated action to limit global average temperature rises to 2°C which financial markets price in gradually)

• Paris Disorderly (same real-world outcomes as the Paris Orderly pathway, but financial markets’ reaction is delayed and abrupt)

• Failed Transition (no additional climate policies are implemented and global average temperature rises by 4°C by 2100).

These are just three plausible pathways, and they are not intended to be “worst case.” We have compared these pathways with a climate-uninformed baseline. This climate-uninformed baseline assumes no increase of physical risks due to climate change and does not make any explicit assumptions about the transition to a low carbon economy.

We have modelled these impacts on the assets and liabilities of an example UK defined benefit pension scheme over the next forty years. The main results of the modelling are as follows:

• The funding risks for the pension scheme are greater under all three climate pathways than under the climate-uninformed base scenario. In the absence of changes to the investment strategy or recovery plan, the time taken to reach full funding is increased by 3 to 9 years.

• For this pension scheme, the worst of these three scenarios is the Paris Disorderly scenario. Although the long-run economic impacts are projected to be greatest under the Failed Transition, the scheme is assumed to have started reducing investment risk by the time the worst effects of the Failed Transition apply. However, the timing of the climate effects is very uncertain, and if the impacts of the Failed Transition started to emerge prior to the risk reductions, then this could be the worst of the three.

As actuaries with some experience in this field, we are satisfied that these results are based on a reasonable model. However, there is material uncertainty in all aspects of climate scenario modelling and this model is not able to fully capture all the risks. For example, it does not incorporate environmental tipping points and is highly dependent on the assumptions used to translate the climate adjustments from GDP onto the other variables. In addition:

• There are some plausible climate pathways (not included in these three) where real gilt yields fall rapidly and so the liabilities could be impacted more extremely.

• Even when the scheme has reduced its investment risks, some climate risks will remain since matching assets are not immune to climate risks (particularly corporate debt due to credit risk) and cannot match uncertain cash flows perfectly.

• Climate risks might affect annuity prices earlier and more significantly than they affect the scheme’s funding position, affecting the cost of transferring the scheme’s liabilities to an insurance company.

• Financial market volatility might increase as the physical and transition impacts of climate change unfold, particularly if this happens in an unpredictable manner. The modelling does not make much allowance for this. For maturing pension schemes, market volatility is likely to increase the chance of being a forced seller of assets and cause a drag on investment returns.

• In all cases, the extent of the sponsoring employer’s exposure to climate risks is relevant: the greater a scheme’s climate risk exposure through the sponsor covenant, the lower its capacity to tolerate climate risk exposure through its assets and liabilities (all else being equal).

These are all areas where further research and model refinements would be useful. In aggregate, it is quite likely that the modelling is biased to underestimate the potential impacts of climate-related risks, especially for the Failed Transition pathway. This reinforces the finding that actuarial advice based on the climate-uninformed baseline would underestimate the funding risks (since the progression of the funding position is worse under all three pathways). Given that most models currently used by actuaries do not make explicit adjustments for climate change, it seems quite likely that pension schemes are systematically underestimating the funding risks they face.

Climate change scenarios can help trustees and employers determine how exposed pension schemes may be to climate change and how the employer may be able to respond as climate impacts emerge. There are then a number of actions that trustees can take in order to mitigate climate risk (not explored in detail in this paper), for example:

• Make changes to investment strategies and their implementation.

• Engage with the employer to understand how resilient it is to climate change and which scenarios it is most exposed to.

• Factor these risks into their funding and investment strategies and plan in advance how to react should they start to materialise.

Finally, the modelling is based on market conditions on 31 December 2019 and makes no allowance for subsequent events, notably the COVID-19 pandemic. The climate impacts we illustrate may seem less significant now they are being considered against that backdrop. However, what COVID-19 does demonstrate is the impact that globally integrated events can have on economies and financial systems and the importance of preparedness.

1. Introduction

Pension schemes are increasingly seeking to understand better and analyse the financial and economic exposure of their assets and liabilities to climate-related risks and opportunities. Climate-informed decision-making is increasingly being demanded by regulators, and for some pension fund trustees, it has become a fundamental consideration as part of their fiduciary duties, to support strategic investment decisions, better risk management and developing more resilience in their portfolios.

This paper explores how climate scenario analysis can be used as a tool for forward-looking assessment of the risks and opportunities for defined benefit pension schemes and other financial institutions. It is intended to provide an example of how pension schemes can integrate climate change into their thinking on matters such as funding strategies and long-term strategic asset allocation and should prove useful to pension actuaries and other advisors who are seeking to give specific consideration to climate risks in their advice to scheme trustees and sponsors. It assumes that the reader already has some knowledge of climate change and its relevance to the work of pensions actuaries, including material covered in other IFoA documents on this topic ¹ . Although the case study that is presented is a defined benefit scheme operating in the UK, this paper should also be useful for actuaries advising other long-term investors, both in the UK and in other jurisdictions. A companion paper “Climate scenario analysis: an illustration of potential long-term economic & financial market impacts” (the Companion Paper), written by the same authors, examines the long-term economic and financial impacts illustrated by the underlying modelling in more detail.

The potential influence and risk of climate change is not in doubt ² . Climate-related risks topped the concerns in the World Economic Forum’s latest Global Risks Report ³ , and almost all countries have signed the Paris Agreement ⁴ , committing them to significant action to limit climate change. It is an unprecedented challenge and poses significant risks to the economy and to the financial system. Some of these risks are already manifesting themselves through changing climatic conditions (such as temperatures and rainfall patterns), rising sea levels and extreme weather events (such as hurricanes, wildfires and flooding). These events have significant negative impacts and disruption on various sectors such as agriculture and forestry, supply chains and infrastructure. Actions to limit the extent of climate change by cutting greenhouse gas emissions, primarily through reducing the use of fossil fuels and switching to renewable energy sources, will fundamentally change many aspects of business and everyday life. Some assets, such as fossil fuel reserves, are likely to lose much of their value if emissions limits are imposed – the problem of stranded assets. The business models in many industries are likely to change, and new industries may emerge. Hence, the socio-economic consequences of climate-related impacts will be wide-ranging and long-lasting. Crucially from an actuarial perspective, these impacts can be difficult to predict, quantify and model due to the systemic nature of climate change.

Climate change will almost certainly fundamentally impact how economies perform as a whole. It will affect macro-economic variables such as GDP growth and in turn have significant influence over the resulting performance of asset classes and industry sectors. Since the risks associated with climate change are systemic in nature, they will affect all assets to some extent and so cannot be avoided completely through careful selection of investments.

These economic factors will affect not only the assets of pension schemes but also their liabilities through impacts on inflation, health and mortality (which may also be affected more directly through the physical impacts of climate change). Moreover, these economic factors will affect the sponsoring employer and its ability to meet any funding shortfalls. Insights gained from climate scenario analysis allow pension schemes and other financial institutions to make more informed strategic decisions with the aim of developing investment portfolios and approaches to liability management that are more climate resilient. This paper considers the economic impacts on the assets and liabilities and hence the funding position. A more comprehensive exercise would also consider impacts on the demographic variables and the sponsor covenant.

Previous IFoA working party papers ⁵ have acknowledged the challenge of assessing the financial impact of climate risks. They have noted the disconnect between existing scientific data describing climate-related risks and the use of this information in traditional financial performance measurement, risk assessment and economic scenario analysis at the strategic level. Moreover, most of the currently publicly available analysis focuses on bottom-up security-specific climate-related risk analysis and reporting and does not consider top-down integration into funding strategies and long-term strategic asset allocation via macro-economic risk analysis.

Meanwhile, there has been an explosion of interest in climate scenario analysis due to its prominence in the TCFD recommendations: “Organizations should describe how resilient their strategies are to climate-related risks and opportunities, taking into consideration a transition to a lower-carbon economy consistent with a 2°C or lower scenario and, where relevant to the organization, scenarios consistent with increased physical climate-related risks.” ⁶ As predicted, we are seeing considerable work in this area as financial institutions start to implement the recommendations. In March 2020, the UK Pensions Climate Risk Industry Group (PCRIG) launched a consultation on its draft guidance for trustees on integrating climate-related risk assessment and management into decision-making and reporting. Climate scenario analysis features prominently in this draft guidance and is described as “a crucial step in trustees meeting their legal duty to manage climate-related risks.” ⁷ Amendments to the Pension Schemes Bill currently before Parliament are expected to make TCFD style reporting – and hence climate scenario analysis – mandatory for large pension schemes ⁸ .

Whilst interest is high, the form, pace and breadth of adoption remain to be seen. We hope that this paper can contribute positively to the pace and breadth of adoption and the emergence of best practice. Pensions actuaries will be able to use the case study to help show them how such analysis can be performed and what the results might show. The case study we present is intended to be realistic, and the numerical results illustrate plausible outcomes rather than forecasts. The climate pathways illustrated are not intended to be extreme and so do not represent “worst-case” scenarios. Moreover, the model we have used does not take account of the full effects of climate change, as it ignores broader environmental tipping points and knock-on effects, such as climate change-related migration and conflict. Given the pathways we have described, there is therefore a strong bias towards optimism in our results.

Climate change is not the only systemic risk to which pension schemes are exposed. Since the majority of the work on this paper was completed, the world has experienced a major shock in the form of the COVID-19 pandemic which is having unprecedented financial market impacts. The climate impacts we illustrate may seem less significant now they are being considered against that backdrop. An important difference is that real-world climate impacts – although already being felt – will manifest themselves over a period of decades, extending beyond the independent lifetime of many UK defined benefit pension schemes, and are likely to emerge much more slowly than those of COVID-19. The mounting awareness of climate risks among financial participants – facilitated by tools such as scenario analysis – should help to mitigate the market impacts by enabling the effects to be priced in gradually, although pricing-in shocks are still likely to occur and could be significant. What COVID-19 does demonstrate is the impact that globally integrated events can have on economies and financial systems and the importance of preparedness.

Climate impacts may be a material factor when considering future insurance buy-out prices (which will need to consider the long-term outlook) as well as nearer term matters such as an appropriate level of deficit repair contributions and the expected time horizon to be fully funded. Moreover, actions taken now will be crucial in determining eventual climate outcomes. For example, atmospheric concentrations of greenhouse gases are determined by cumulative emissions, hence the level of emissions today matters as well as the level at future dates (e.g. 2050), and infrastructure being built now could either lock us into fossil fuel dependency for decades or accelerate the transition to a low carbon economy.

2. The Systemic Nature of Climate Risk

As outlined in section 1, climate change is an unprecedented challenge and poses significant risks to the economy and to the financial system. The risks are often classified into two main types:

• Physical – these arise from both gradual changes in climatic conditions and extreme weather events; and

• Transition – these arise from the move to a low carbon economy and include impacts relating to policy, technology, markets and reputation ⁹ .

There are likely to be significant opportunities too, particularly in relation to the transition, although the net impact is likely to be negative for many sectors and for the financial system as a whole.

The physical and transition impacts that actually materialise will depend crucially on the speed and magnitude of climate change and the policy response. There are many possible pathways, driven by the choices made by governments and society at large, varying from those in which our behaviour does not change (leading to more severe physical impacts) to those in which drastic action is taken to counteract climate change (leading to more severe transition impacts). The actual pathway that we will follow is highly uncertain. Nonetheless, it is certain that there will be financial impacts from some combination of physical and transition effects.

These climate impacts will have macroeconomic consequences (i.e. economy-wide effects) that will affect indicators such as GDP growth, interest rates, inflation, investment and international trade flows. This will in turn affect risk-return expectations across all asset classes, regions and sectors and also the value placed on the liabilities of pension schemes. The risks are therefore systemic in nature – they will affect large parts of the economic and financial system, with many companies potentially affected in similar ways at similar times. The Bank of England has warned repeatedly that climate risks could cause instability of the financial system ¹⁰ .

It is therefore imperative that pension schemes and other financial institutions start to assess and determine the potential impact of transition and physical risks on their assets and liabilities. Not only will this help individual schemes navigate the uncertain waters ahead, it will also help facilitate an understanding of how, and on what scale, climate change creates risks for the whole financial system.

2.1 Modelling the Financial Impacts of Climate Change

The modelling of climate systems is a long-established area of research. Considerable work has also been carried out to model the economic costs and benefits of climate change and related policy action. More recently, attention has turned to modelling the effects on financial markets and financial institutions. However, most models currently used by actuaries do not make explicit adjustments for climate change. They often assume that market prices fully reflect climate-related risks, despite the warnings of financial regulators and others that this is unlikely to be the case.

Of the analysis of financial impacts that is currently available, most focuses on bottom-up climate-related risk analysis and reporting of investment impacts and does not consider top-down integration via macro-economic risk analysis into asset-liability modelling and long-term strategic asset allocation.

Although bottom-up approaches for carrying out climate scenario analysis are useful for some purposes, they suffer from a number of disadvantages. They are mainly focused on individual companies or sectors on the assets side or individual effects on health or mortality on the liabilities side. They are unable to take economic networked effects into account, thus missing the structural impacts on the global economy and making it difficult to consider assets and liabilities consistently.

A top-down or “systemic” perspective is increasingly being considered in the academic literature, as well as in guidance documents, and by regulators and key experts in the field ¹¹ . Such a top-down approach to scenario analysis captures the systemic nature of climate risks that are not identified in “bottom-up” metrics and methodologies.

It is acknowledged that a micro-economic approach is necessary to drill down to the individual holding level. This is important for understanding how climate risks are distributed within the investment portfolio and hence where mitigating actions should be focused. Therefore, “top-down” and “bottom-up” approaches to climate scenario analysis are complementary and both need to be considered.

3. Introducing the Systemic Climate Risk Scenario Tool

For this paper, we have used ClimateMAPS, the top-down modelling tool developed by Ortec Finance in partnership with Cambridge Econometrics. This tool combines climate science with macro-economic and financial modelling in order to integrate quantified systemic climate risks and opportunities (both physical and transition) into traditional multi-horizon real-world scenario sets. The climate risk integration logic “ties together” climate science, macro-economic modelling and financial modelling. The methodology is relevant for a broad range of financial organisations, covers both physical and transition risks and provides granularity down to the country and sector level across asset classes for several global warming pathways and stress-test scenarios. It is described in more detail in Appendix A.

ClimateMAPS is not the only modelling tool that could have been used. The precise numerical results from the modelling would be different had another tool been used. We consider that ClimateMAPS provides useful insights in spite of its limitations (see section 3.2), and think it likely that other tools might provide similar insights – an interesting area of further work would be to perform some comparisons across different modelling tools.

We have used ClimateMAPS to model the effects of three different climate pathways through sets of alternative economic assumptions that mimic various possible evolutions of the economy depending on which climate pathway the world follows. The scenarios aim to be a realistic representation of possible policy, technology and physical risk developments under different temperature pathways. They therefore represent plausible climate-aware real-world scenarios, rather than climate stress-test scenarios, i.e. they are not potential worst-case scenarios.

Each pathway includes a set of climate-related risks and opportunities. These risks can be further broken down into physical and transition risks. For example, risks arising from a transition to a low-carbon economy require a large shift of investments across all sectors towards low-carbon alternatives. On the other hand, if the world continues on an unchanged path, more physical risks will come into play and disrupt expected growth. It should be noted that there are some physical effects that will materialise regardless of the pathway, due to locked-in impacts caused by greenhouse gases already emitted over the past decades. These physical effects are much smaller if we transition towards a low-carbon economy, but they should not be neglected.

3.1 Climate Pathways

The three climate pathways that we have modelled are shown in Figure 1 and described in more detail in Appendix B. They consist of two pathways in which the world transitions to a low-carbon economy in a way that is consistent with the Paris Agreement, and one in which there is no such transition.

Figure 1. Climate change pathways modelled ¹² .

In the Paris Orderly Transition pathway, action to achieve the Paris Agreement goals starts immediately and continues until net zero global emissions are achieved in 2066. Because of the timely awareness and response of policy and financial actors, the pricing in of climate-related risks (transition and physical) takes place gradually over the period 2020–2024. It should be noted that this pathway is expected to limit average global temperatures to below 2°C (75% probability) but not necessarily 1.5°C (50% probability) by the end of this century. The alternative 1.5°C narrative is of increasing interest to financial institutions, especially those which have committed to net-zero emissions by 2050. Achieving this pathway with a greater probability than 50% would require an even steeper transition pathway (based on IPCC SR1.5 instead of RCP 2.6) with quicker technology take-up, stricter policy measures and negative emissions from carbon capture technologies and/or afforestation/reforestation measures.

The Paris Disorderly Transition pathway includes the same real-world transition risks and opportunities as the Paris Orderly Transition pathway. The assumptions regarding matters such as policies across sectors and countries as well as technology uptake are the same in both pathways. The physical risks associated with a Paris pathway, i.e. those gradual physical risks and extreme weather events associated with staying below 2°C global average temperature rise from pre-industrial levels ¹³ by 2100, are also the same for both. What differs, however, is how markets become aware of, and address, these transition and physical risks. In the Paris Disorderly Transition pathway, the financial markets have little initial awareness of the scale and speed of the transition that is required. There is an abrupt realisation of the problem in 2024, followed by a re-evaluation of assets directly and indirectly related to carbon-intensive economic activities. In addition, stocks and bonds are abruptly re-priced in 2024, with a consequential sentiment shock, i.e. a carbon bubble/stranded asset event, in 2025. These abrupt responses on financial markets also lead to increased market volatility from 2024 to 2026. ¹⁴

In the Failed Transition pathway, there is no impetus for policymakers to implement additional policies over and above those already in place, and therefore, the Paris Agreement goals are not achieved. The scale and potential impact of the physical risks to 2050 that this pathway implies are priced in over the period 2025–2029. The severe physical risks beyond 2050, and the structurally lower growth expectations due to these physical risks, are then considered as part of a second repricing in the period 2035–2039.

These three climate-aware pathways are compared to a climate-uninformed baseline economic outlook, in which no account is taken of the effects of future temperature increases, climate-related policies or technology trends. More details of the baseline are provided in Appendix A.

3.2 Limitations

Given that the future is uncertain, the random variation in future economic variables and investment returns over the short term may result in experience that is significantly different to the expected long-term average experience. This is true of all stochastic financial models but is particularly important here because there is material uncertainty in all aspects of climate scenario modelling. The use of judgment is required at all stages in both the formulation and application of climate scenario models.

Furthermore, as most current economic models are built on past experience, including implicitly the assumptions that climate change is not occurring, and there is no energy transition, there is an important question for users of these models about the level of uncertainty inherent in them.

The modelling is intended to illustrate plausible impacts on a hypothetical UK defined benefit pension scheme. It considers the economic impacts on the assets and liabilities, and hence the funding position, but not the impacts on demographic variables or the employer covenant. The impacts will vary significantly by scheme particularly because of differences in (amongst other things) time horizon, risk appetite, employer covenant, investment strategy and funding position. Furthermore, rather than focusing on the absolute results under each pathway, we encourage readers to focus on the relative results of the climate-aware pathways compared with the climate-uninformed baseline.

The pathways we have modelled do not cover the full range of possibilities. For example, our Paris Disorderly Transition assumes a late realisation only on the part of the financial markets of the physical and transition risks of climate change. Other possible drivers of disorder include a late realisation of the risks by policymakers leading to abrupt policy action, unexpected technological breakthroughs, or a sudden shift in consumer sentiment. These would all result in disorderly impacts that would differ in impact and timing from what we have modelled. Moreover, the actual outcome is likely to be different from any of our pathways. However, the pathways do give some idea of the types of impacts that may be seen and of their potential relative significance.

The model we have used relies on Cambridge Econometrics’ macro-econometric model E3ME to integrate transition and physical risk drivers and calculate their impact on macro-economic outputs (see Appendix A). E3ME considers only carbon dioxide (CO₂) emissions from energy use and does not model other emissions or emissions from agriculture or changes in land use. In order to capture the effects of other greenhouse gas emissions, the model uses a climate sensitivity coefficient that implicitly includes these other emissions. In addition, E3ME assumes that the supply of natural resources (e.g. water, forest) is equal to demand, i.e. only demand is modelled, not supply. Other models are available but as actuaries with some experience in this field, we are satisfied that this is a reasonable model and captures most of the features that we would like.

The modelling translates the impacts of climate-adjusted GDP shocks onto a wide range of financial and economic variables. To do this, GDP is the only translation mechanism from the macro-econometric model to the stochastic financial scenario model. Other potential translation mechanisms (such as carbon-price impact on inflation and interest rates) are out of scope and follow purely from the estimated relationships with GDP in the financial model. The results of the modelling are highly dependent on the assumptions used to translate the GDP shocks onto the other variables.

There is particular uncertainty about how climate change might affect interest rates and inflation since there has not yet been much research in this area, and the available evidence is mixed. Historically, inflation and interest rates have generally been lower when economic growth is low. In this model, inflation and interest rates fall broadly together in the climate pathways which means that the real interest rate, which is the most important driver of pension scheme liabilities, does not change that much. That is, in the coming 20 years, real rates are assumed to decline by 10 (Paris) to 15 (Failed Transition) basis points relative to the baseline. By 2060, real rates are expected to be 30 (Paris) to 60 (Failed Transition) basis points lower than the climate uninformed baseline. However, plausible narratives can be constructed in which interest rates fall but inflation is stable or even rises. As an example, Aon has produced a “Forced Green” scenario in their Climate Change Challenges paper¹⁵ which explores the impact of delayed action for five years, with governments eventually forced to address greenhouse gas emissions due to increasing extreme weather events leading to lower nominal yields but higher inflation. Such scenarios, which are not considered in this paper, could lead to significant increases in the value of liabilities.

Existing research on how climate change affects financial market volatility is limited and inconclusive. Volatility might increase as the physical and transition impacts of climate change unfold, particularly if this happens in an unpredictable manner. Due to the inconclusiveness of the research, the modelling does not make any allowance for this, except in the Paris Disorderly Transition pathway during the period 2024–2026 whilst pricing in of climate-related risks takes place.

The outputs of this paper focus on median results and compare the projections to a “climate-uninformed” baseline that assumes historic trends with no allowance for additional climate impacts. Almost all the charts and tables are referenced to this baseline, so that calibration of the baseline largely cancels out, enabling the discussion to focus on the economic and financial impact of the three different climate pathways. The impact on tail risks will be significantly influenced by any increases in financial market volatility, and thus, we have focused on median results in this paper. Future work could be undertaken to consider the impact of different baseline assumptions and further investigation of tail risk impacts.

There is a great deal of subjectivity in the assumed timing of market movements. In our model, the biggest market movements under a Failed Transition pathway occur after 2030, which is after our example scheme has started reducing the investment risks. The worst impacts are thus avoided. However, the market movements could occur a lot earlier, in which case the funding position would look a lot worse. For example, analysis by the Cambridge Institute for Sustainability Leadership of a “No Mitigation” scenario found that all portfolios experience losses over the next five years, the largest being a 45% loss for the portfolio with the highest equity allocation (60%)¹⁶.

The modelling does not incorporate any changes to the definitions of UK inflation measures, for example in the gap between RPI and CPI measures. The effects of any changes are expected to be very similar under each scenario, so there would be minimal impact on the gap between the climate-aware scenarios and the climate-uninformed baseline.

The modelling is based on market conditions on 31 December 2019 and makes no allowance for subsequent events, notably the COVID-19 pandemic.

Furthermore, it should be noted that the modelling does not consider broader environmental tipping points and knock-on effects, such as climate change-related migration and conflicts. Nor does it consider the potential for food or other resource shortages which may lead to both lower GDP and higher inflation. In aggregate, it is quite likely that the modelling is biased to underestimate the potential impacts of climate-related risks, especially for the Failed Transition pathway.

4. Case Study: Applying the Tool to a UK DB Pension Scheme

The case study in this section is based on a hypothetical UK defined benefit pension scheme and provides an illustrative example of the type of forward-looking, quantified analysis that can be done using climate scenario analysis. It provides a perspective of climate-related risk under three different global warming pathways: a Paris Orderly Transition, a Paris Disorderly Transition and a Failed Transition. All three climate pathways are compared to a climate-uninformed baseline.

Due to the long-term nature of climate risks, the time horizon is longer than typical periods utilised by pension schemes. The case study analysis covers projections over 40 years (until 2060) and is either expressed in absolute levels or as a difference to the baseline (climate-uninformed) pathway, i.e. an unadjusted econometric model. The baseline and all climate-adjusted economies used in this case study are built on the December 2019 market situation.¹⁷

The key characteristics of the case study are given in section 4.1, with more details of the investment profile in Appendix C. Section 4.2 describes the progression of the scheme’s funding position under the three climate pathways. Appendix D outlines the GDP and asset class returns underlying the pathways.

4.1 Scheme Description

The example pension scheme that we use in our case study has been chosen to be fairly representative of the current situation of many DB schemes in the UK ¹⁸ . It has a relatively high initial allocation to growth assets and relatively low hedging ratios which results in higher exposure to climate risks than a more mature scheme with a low allocation to growth assets. It retains its exposure to growth assets throughout the next decade. The scheme has the following characteristics:

• Liability profile. 80% of the cash flows are linked to RPI, and the remainder are fixed, with an initial duration of around 17 years. There is no further accrual of benefits. The projections focus on a Long-Term Funding Objective, calculated using a discount rate of gilts+0.5% pa ¹⁹ . On this basis, the scheme starts with an initial value of liabilities of £133 million.

• Investment profile. The initial market value of assets is £100 million, which are initially invested 50% in growth assets and 50% in bonds, with around 50% of the liabilities hedged against interest rate and inflation risk. From 2030 to 2040, it is assumed that the scheme will linearly reduce the risk to be 100% bonds with 100% interest rate and inflation hedging. More details are included in Appendix C.

• Cash contributions. The liabilities and investments give a starting funding level of 75% on the Long-Term Funding Objective. To help to eliminate the deficit, a Recovery Plan has been agreed such that company contributions of £670,000 are paid annually over the first 10 years. These contributions have been set based on maximum affordability to the sponsoring company, such that they do not impact on the company’s growth plans.

On non-climate adjusted median assumptions (by which we mean that there is a 50% chance that experience will be better and a 50% chance that experience will be worse), we estimate that the scheme will reach 100% funding on the Long-Term Funding Objective after 10 years (assuming the Recovery Plan is followed in full). This is shown by the median baseline case in Figure 2.

Figure 2. Funding level projections: 5th, 50th and 95th percentile outcomes.

It is assumed that no changes are made to cash contributions or the asset allocation in the light of experience.

4.2 Pension Scheme Performance

In this section, we analyse the performance of the pension scheme across our three climate-informed pathways and the climate-uninformed baseline. For this, we generate 2000 stochastic economic scenarios for the period 2020–2060 for each climate pathway ²⁰ . For each pathway for each scenario for each future year within this time period, we calculate the investment return, the resulting value of the assets and the value of the liabilities (taking into account inflation and interest rates in the specific scenario), and the resulting funded ratio. In this section, we show the impact of the different pathways on the funded ratio and discuss some of the considerations that apply. Appendix D analyses underlying drivers of the differences in impact that we see.

Each of the pathways is intended to be plausible, but we have not attempted to assign a probability to the likelihood of each one occurring. We do not know which climate pathway will transpire in practice – and it could look quite different to those we have modelled – so the outcomes illustrated in Figure 2 can best be interpreted as widening the funnel of doubt that pension schemes should consider. In other words, rather than considering the 5th to 95th percentile range for one scenario, readers should consider the range from the worst 5th percentile case to the best 95th percentile case and recognise that this is likely to represent much less than the middle 90% of possible climate-aware outcomes given the uncertainties involved.

Figure 2 shows the following impacts on progress towards the Long-Term Funding Objective:

Figure 3 illustrates the median outcomes of the funding level projections: focus on median outcomes when they are depicted as difference to climate-uninformed baseline. We can see that the climate-uninformed baseline is too optimistic and underestimates the funding risks for this pension scheme. Furthermore, the Long-Term Funding Objective is met much later in the Paris Disorderly Transition pathway than in the other pathways. The probability that this pension scheme will be fully funded on this basis at the originally-expected date of 2030 is less than 50% unless additional deficit contributions are paid.

The differences in funding level development between the different pathways are mainly due to differences in investment returns up to 2030 (see Appendix D). The analysis assumes that the scheme will not make any changes to its asset allocation between 2020 and 2030 and that contributions will be payable throughout this period. After 2030, the pension scheme reduces the investment risk to one with 100% invested in fixed income, and 100% hedging of inflation, interest rate and currency risk. The funding level trajectory beyond 2030 is therefore predominantly determined by the funding level in 2030. However:

• If the funding were to be below target in 2030, then the trustees may decide not to reduce the investment risk but to target the same returns as before. In this case, the situation is likely to look worse for the Failed Transition pathway as the growth asset markets suffer more losses after 2030. The box below illustrates what might happen if the scheme did not reduce the investment risks during the 2030s.

• If the contributions were already at their maximum affordability, but the company then has to divert more of its cash in dealing with impacts from climate change, then it may be looking to reduce its cash contributions into the pension scheme at future valuations. This could be the case even in the Failed Transition and particularly under either Paris Transition pathway.

• The Paris Orderly Transition pathway shows the most favourable climate-informed median funding level results in the medium to long term. This pathway is also far preferable for the economy as a whole as this pathway is the least disruptive in the medium to long term.

In interpreting these results, it is important to note that it is quite likely that the modelling is biased to underestimate the potential impacts of climate-related risks, especially for the Failed Transition pathway. In addition, readers should look at the wider context rather than considering these results in isolation. Under the Failed Transition pathway, catastrophic physical impacts would be expected in the second half of this century with severe consequences for the overall wellbeing of society. Hence, whilst the modelling above suggests that, from the scheme’s perspective, outcomes under the Paris Orderly Transition and Failed Transition pathways are similar, from a societal perspective even the Paris Disorderly Transition is far preferable to a Failed Transition. The differing long-term impacts and the attributions to different elements of climate risks are discussed further in the Companion Paper.

Box 1. What-if analysis without the reduction in investment risk

One of our key modelling assumptions is that our hypothetical scheme follows a reducing investment risk path that reallocates the exposure to equities and real estate to fixed income assets during the 2030s. This assumption benefits our scheme as it hedges some of the climate risks that would occur in the second half of the time horizon. Indeed, the second pricing-in shock modelled in the Failed Transition pathway mostly does not affect our scheme’s performance thanks to this reallocation of assets. However, what would happen if our scheme keeps its asset allocation static? Figures 4 and 5 below present the evolution of the funded ratio under such an alternative static asset allocation.

Comparing these results with the first scenario, we can see that the scheme performs better on average if it does not reduce investment risk. However, the persistent allocation to equities leads to a higher risk level. In particular, this alternative scheme performs worse under the Failed Transition pathway. As the second pricing-in shock occurs, the gap between the Paris Orderly Transition and the Failed Transition pathways widens to the point where the Failed Transition falls below the Paris Disorderly Transition in 2038.

Figure 3. Funding level projections: focus on median outcomes (depicted as difference to climate-uninformed baseline).

Note: Difference to baseline is calculated as the ratio of climate-informed median and baseline at each year.

Figure 4. Funded ratio projections under a static asset allocation.

Figure 5. Funding level projections under a static asset allocation: focus on median outcomes (depicted as difference to climate-uninformed baseline).

Note: Difference to baseline is calculated as the ratio of climate-informed median and baseline at each year.

We have summarised further risk metrics related to the funded ratio in Table 1 below. We see that using the climate-uninformed baseline strongly underestimates the funding risks of this pension scheme. Especially in the Paris Disorderly Transition pathway, the long-term probability of underfunding is significantly higher than the probabilities emerging from the climate-uninformed baseline. Of the climate-informed pathways, the Paris Orderly Transition pathway is the most favourable in the long run for our pension scheme.

Table 1. Selected metrics for the funding level (percentages)

The following factors should also be considered alongside the modelling:

• Even once the scheme has reached full funding on a gilts+0.5% basis, there remain risks which we have not captured. The trustees may have a further target to buy-out the benefits with an insurance company. We have not considered how buy-out terms may be affected by climate change but the longer buy-out is delayed, the more uncertainty there is likely to be.

• The pension scheme is maturing. This means that cash outgoings will increase as a proportion of assets. In this environment, volatility in the markets is likely to increase the chance of being a forced seller of assets and cause a greater drag on investment returns. Climate change is likely to lead to greater volatility. As noted in section 3.2, the modelling incorporates greater volatility in the Paris Disorderly Transition pathway, but not the other climate-aware pathways.

• We have assumed a constant level of contributions. The strength of the sponsoring company covenant may also be impacted by climate change, and this should be considered alongside the funding impacts if possible.

• The principal limitations of our modelling are described in section 3.2. These are expected to underestimate the potential financial impacts of climate-related risks.

5. Conclusion

There is an increasing expectation that pension schemes and other financial institutions use scenario analysis to understand their potential exposure to climate-related impacts. As noted in section 1, climate scenario analysis is expected to become a legal requirement for large UK pension schemes soon. This paper illustrates one possible approach that uses a top-down model to explore the financial impacts of three plausible climate pathways on the funding position of an example UK defined benefit pension scheme. The progression of the funding position is worse under all three pathways, indicating that advice based on the climate-uninformed baseline would underestimate the funding risks. Given that most models currently used by actuaries do not make explicit adjustments for climate change, it seems quite likely that pension schemes are systematically underestimating the funding risks they face. In the UK, this may be a concern for the Pensions Regulator and may also have implications for the Pension Protection Fund.

It is important to note that the climate pathways illustrated are not intended to be extreme, even though they can result in extreme financial outcomes as illustrated by the 5^th and 95^th percentiles. The climate impacts for the example scheme under “worst-case” scenarios could be much larger than those illustrated. Moreover, the model we have used does not take account of the full effects of climate change, so there is a strong bias towards optimism in our results.

The example scheme is closed to accrual, so the bulk of cash flows occur in the next 25 years. Due to the reduction of investment risk that occurs from 2030 onwards, the scheme is most exposed to climate risks prior to this date. We expect this time horizon to be typical for DB pensions actuaries’ work, but some schemes – for example, those which are less mature and will remain invested in growth assets for longer – will have greater climate risk exposure. Conversely, schemes that have already largely reduced their investment risks may have fairly limited exposure to climate risks. Nonetheless, some climate risks will remain since matching assets are not immune to climate risks (particularly corporate debt due to credit risk) and cannot match uncertain cash flows perfectly. In all cases, the extent of the sponsoring employer’s exposure to climate risks is relevant: the greater a scheme’s climate risk exposure through the sponsor covenant, the lower its capacity to tolerate climate risk exposure through its assets and liabilities (all else being equal).

There will be significant timing risk for many schemes – outcomes will depend on precisely when climate shocks and market repricing occur relative to the risk reduction path as this could disrupt a scheme’s journey planning. Particular care should be taken when modelling “disorderly” pathways in schemes with significant investment risk reduction expected over the next few years so that the scenario shock is still considered and not arbitrarily assumed to occur after most of the risk reductions have taken place.

If schemes expect to transfer their liabilities to an insurance company, they need to consider how climate risks might affect insurer pricing in future (or even the solvency of the insurer). Climate risks might affect annuity prices earlier and more significantly than they affect the scheme’s funding position. Indeed, we anticipate climate change will receive increasing attention from UK life insurers in the next few years due to the Prudential Regulatory Authority’s recent supervisory statement on climate change ²¹ and the Bank of England’s proposed climate stress tests ²² . Hence, the longer-term financial impacts of climate risk may be relevant even for pension schemes that have already largely reduced their risks.

For actuaries advising pension schemes, it is clear that some investment strategies are likely to be materially impacted under climate pathways relative to climate-uninformed projections. In these circumstances, analysing the potential impacts can inform actuaries’ advice on likely time horizons to full funding, impacts of different investment strategies and suitable deficit repair contributions. Managing the trade-offs for climate risks is particularly challenging as the traditional counterbalances to an increased deficit of larger allocation to growth assets or longer time horizons may both increase sensitivity to climate-related risks and may be no longer viable. Such tensions may be eased by considering the bottom-up perspective that was mentioned in section 2.2. The bottom-up perspective can help build investment portfolios that are more resilient to climate risks than the traditional market cap benchmark.

Climate scenario analysis for financial institutions is a relatively new area where significant work is currently underway. Over the next few years, we expect modelling approaches to become more sophisticated and consensus to start to emerge around which scenarios to use. In the meantime, we hope this paper is helpful to actuaries and others in demonstrating how such analysis can be performed and what the results might show. We encourage them to consider whether their current models may be underestimating the risks from climate change and, if so, work to address this.

Trustees and employers can use climate-change scenarios to help determine how exposed they may be to climate change. This includes factoring in how the employer may be able to respond in the scenarios – for example whether they would be able to afford an increase in contributions or even whether they would be able to continue the current level of contributions. Historically trustees have been nervous about taking action which may be deemed to not be in the best financial interests of their members. However, scenario modelling can help give the trustees the evidence they require in order to take actions to help mitigate climate change risks and it is rapidly becoming the case that trustees will be deemed to be not acting responsibly if they do not take any action. Indeed, there are a number of actions that trustees can take in order to mitigate climate risk, for example:

• There are many changes which can be made to investment strategies, which are beyond the scope of the paper:

• They can engage with the employer to understand how resilient it is to climate change and which scenarios it is most exposed to. This may encourage the employer to take steps to reduce these risks.

• In any event, the trustees can at least factor these risks into their funding and investment strategies and plan in advance how they would react should they start to materialise. In this way, they can hopefully mitigate at least some of the risks.