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, United Nations Principles for Responsible Investment (PRI) (2019), Institutional Investors Group on Climate Change (IIGCC) (2019). 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.

3.1. 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, United Nations Framework Convention on Climate Change (UNFCCC) (2015), and one in which there is no such transition.

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 Orderly 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 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 Orderly 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 repriced 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.

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 judgement 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 possible plausible impacts on the economy and financial markets. 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, forests) 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 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 paper, Aon (2018), 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 inflation-linked 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 Orderly Transition Pathway during the period 2024–2026 while 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. Those looking to mitigate the impact of the risks described should be wary of the timing risk they may be taking on and in particular the potential for an output to flatter a particular approach which is highly dependent on the timing within a model, if this is relatively arbitrarily assumed.

The modelling does not incorporate any changes to the definitions of UK inflation measures, for example in the gap between retail price index and consumer price index (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 at 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.

4.1. GDP Impact: UK and World to 2100

We examine the impact of the climate pathways by first considering the impact on GDP expressed as a percentage of baseline GDP where there is no explicit allowance for climate change as shown in Figure 2.

In the UK, the initial impacts on GDP in the Paris pathways are fairly limited, with positive stimulus effects of low-carbon investment offset by physical impacts and decommissioning carbon intense activities, as shown in Figure 2. In the longer-term, investments in low-carbon electricity and energy efficiency improvements are paid back, resulting in lower levels of consumer spending. Although the UK benefits from the recycling of carbon tax revenues (through reduced VAT and income taxes), this effect is not sufficient to outweigh the negative impacts of physical risks, higher prices for consumers and debt repayment, driving an overall small negative transition impact on GDP in the long-term.

The impact on UK GDP from the Failed Transition Pathway is initially similar to the Paris Transition pathways. However, the higher impact of physical damages is noticeable even by 2040 and continues to accelerate as the physical impacts increase, resulting in GDP which is 60% lower than baseline by 2100.

Ortec Finance’s modelling indicates that over the longer-term the UK is expected to experience larger GDP impacts from the physical effects of climate change than the European average, in particular under the Failed Transition Pathway. This is because many UK cities have relatively higher exposure to extreme weather event risk as they are at higher risk of more severe flooding.

The World GDP projections are similar to the UK, albeit GDP declines further to 20% below baseline by 2100 in the Paris Pathways and is 73% below baseline in the Failed Transition Pathway, reflecting the higher physical impacts on the world as a whole.

Figure 2 illustrates the long-term impact that climate change can have on GDP and in particular how the physical impacts of a Failed Transition could damage the long-term economy. Relative to no climate change, the median projected long-term impacts of Paris transition are very modest. For example, the total fall in UK GDP (relative to baseline) under the Paris pathways is 11.5% to 2100 which is 0.15% p.a. In contrast, the median projected drop in UK GDP to 2100 under the Failed Transition Pathway is 60%, which is 1.1% p.a. It is notable that in the Failed Transition Pathway GDP tracks the Paris pathways fairly closely in early years, but then climate impacts accelerate through to 2100.

Further details on the decomposition of GDP impacts by type of climate risk (transition, gradual physical, extreme weather and financial market sentiment) are given in Appendix C.

4.1.1. Nominal GDP growth under climate pathways: UK and World

The charts of nominal GDP show that, in the long run, unmitigated climate impacts (Failed Transition) are projected to overwhelm the intrinsic growth of GDP which starts to fall in both the UK and globally, as shown in Figure 4. The damage and disruption caused by climate change overwhelm the capacity for productivity and innovation gains. This would create a fundamental challenge for modern financial markets which have no precedent for long-term falls in world GDP.

Figure 4. Climate-adjusted GDP UK and World growth (2020 = 100).

As noted in Chapter 3.2, these projections make no allowance for geopolitical tensions, migration and food supply disruption which are likely to significantly add to the negative impacts. And finally, these charts show median outcomes.

On the other hand, a low-carbon transition that achieves the Paris Agreement targets, especially if done in an orderly manner, does not have noticeable negative consequences on GDP until mid-century. The lower growth expectations even under a successful transition are primarily explained by two factors: the repayment of debt that helped finance the transition; as well as the increasing impacts of physical risks in a + 2°C world compared to what we experience today (approximately 1.1°C world). Even the later years’ impacts remain modest at approximately 0.3% p.a. lower world GDP growth over the period 2020–2100.

4.1.2. Regional GDP impacts

In studying the GDP impacts, it should be noted that there can be considerable variation in the impacts across geographic regions. These differences are mainly caused by the level of transition risks over the short-term and physical risks over the longer-term. Examples are given in Figure 5 which shows the differences in GDP impacts across four countries (Canada, China, UK and US).

Figure 5. Climate-adjusted GDP growth across regions and climate pathways (ratio of cumulative medians to the cumulative median of climate-uninformed baseline pathway).

In a Paris transition, Canada GDP takes an initial hit due to the steeply decreasing demand for its tar sands and other energy-intensive fossil fuel extraction. Longer term it recovers almost all of the losses due to its other natural resources and the ameliorating climate being a net positive with 2 degree warming. However, in a Failed Transition, whilst the short-term impacts are neutral, in the longer term the negative physical climate impacts of 4 degrees warming on the Canadian and the world economy overwhelm the initial ameliorating climate effect.

Similar to the UK, the initial Paris pathways in China are similar to baseline. But China is more sensitive to longer-term physical risks which means greater longer-term GDP impacts with declines by 2100 of 20% for Paris-aligned pathways and 70% for the Failed Transition relative to baseline.

In the US, there is a slight initial net GDP loss under the Paris-aligned pathways reflecting reductions in shale fossil fuel and coal production, but it is not nearly as severe as Canada. The longer-term physical impacts of the Paris-aligned pathways are approximately 15% of GDP by 2100, i.e. between those of UK and China. US is also between their sensitivities to the physical risks in the Failed Transition Pathway which drops approximately 65% by 2100 relative to a climate-uninformed baseline.

As these examples illustrate, transition and physical impacts are likely to play out rather differently in different countries around the world. Thus, to consider the long-term impacts of climate change it is important to consider these country-level insights into systemic climate risks in addition to a micro-level analysis of individual assets.

4.2. CPI Impact

In the current modelling set-up and assumptions, CPI is not shocked separately, but is affected via the historic relationship with GDP. A reduction in long-term economic growth is assumed to lead to a reduction in inflation in the long term, as has occurred in previous recessions. However, it might be the case that certain climate impacts cause a negative supply shock, particularly on food, and in this case one could expect CPI to increase in the short term, although this potential impact is not modelled.

The modelling of CPI, and all financial market impacts, stops at 2060, as shown in Figure 6. This is because the relationships with unprecedented downward-trending GDP cannot be reliably determined. Furthermore, other variables such as agricultural production and inflationary impacts of supply chain disruption are not modelled and are likely to be material factors for inflation in the second half of the century.

Figure 6. Climate-adjusted CPI (annualised difference of medians to median climate-uninformed baseline pathway).

Within the Paris Orderly Pathway, the median modelled UK CPI impacts relative to the climate-uninformed baseline are very modest, less than 0.1% p.a. until 2050 when they decline towards 0.3% p.a. lower than the baseline. The initial slight rise is due to the higher investment demand for the low-carbon transition. It then declines as GDP declines, with fractional interactions between investment, debt repayment and physical damage. Longer term, CPI is assumed to drop to approximately 0.3% p.a. lower than the baseline, reflecting the fractionally lower GDP growth.

The transition shock of the Paris Disorderly pathway creates short-term inflation volatility, but inflation then settles along the Paris Orderly level from 2035 onwards.

In the Failed Transition Pathway, CPI drops to 0.5% p.a. lower than baseline by 2060, reflecting the lower GDP growth. As noted above, other factors that have not been modelled may increase rates of inflation although higher costs of production would reduce GDP further than projected by the model.

4.3. Nominal Interest Rate Impact

The median of the UK 20-year government interest rate is not affected much in the short-term due to limited GDP impacts across all the pathways, as shown in Figure 7. In the mid-term, the UK experiences some negative impacts from higher debt due to the transition. In the longer term, the UK gilt yield declines due to lower GDP forecasts. This long-term fall is relatively modest in the Paris pathways (0.4% to baseline by 2060 which has an assumed nominal interest rate of 4.2%). However, the significant drop in GDP for the Failed Transition Pathway leads to a more significant (1%) drop in yields in 2060 relative to the baseline. 20-year UK gilt yields are relatively insulated within the Paris Disorderly pathway. The impact of a disorderly shock is assumed to be relatively modest with an initial yield rise, some volatility and then returning to the same path.

Figure 7. Climate-adjusted 20 year nominal yields of UK government and investment grade bonds (annualised difference of medians to median climate-uninformed baseline pathway).

Investment grade bond behaviour for the Paris Orderly and Failed Transition Pathways is assumed to be quite similar to sovereign bonds, although the Failed Transition effect is slightly smaller in the long term (0.8% drop versus 1%) because spread levels are expected to increase due to increasing physical risks. However, the market disruptions, within both the Paris Disorderly and Failed Transition Pathways, are more pronounced for investment grade bonds, leading to greater volatility.

5.1. Paris Orderly

In the Paris Orderly Transition Pathway, we assume that pricing-in will take place in the coming five years, i.e. 2020–2024. During this period, equity returns are impacted by the impact on earnings of the transition costs and a repricing due to the lower long-term future expected earnings and dividends in fossil-fuel-intensive sectors.

Future physical risks explain most of the lower long-term returns and thus the repricing in the short-term. Long term, there is a modest net repayment for the transition investments. The majority of the longer-term impact is then addressing the cost of the gradual physical risks as they arise. The pricing-in shock in 2020–2024 only takes into account impacts until 2050, so physical impacts beyond that date are not priced in yet, impacting markets only at a later stage. Furthermore, although we assume that financial markets have a general understanding of the long-term impact of physical risks, and this is priced-in in the coming years, we do not assume perfect foresight. Because the exact timing of physical risks and extreme weather events materialising is uncertain, they still have some impact on financial markets when they actually occur. The overall long-term physical impacts remain modest at approximately 0.5% p.a.

5.2. Paris Disorderly

In the Paris Orderly Transition Pathway, we assume that pricing-in starts later, in 2024, and that the full price correction materialises within one year. This large price correction leads to negative market sentiment, and consequently to a sentiment shock in 2025, followed by (positive) price corrections in consecutive years to compensate for the overshooting in previous years. Market volatility is extremely high in this 2024–2026 period. Over the long-term, the impacts are the same as Paris Orderly but not directly visible on this chart due to the scale.

5.3. Failed Transition

The initial impacts of the Failed Transition Pathway are negligible as the current physical impacts of climate change have little global economic impact. There are no additional transition costs relative to those required in current climate policies, which are included in the baseline. However, in the second half of this decade, we assume that investors will start to realise that the transition to a low-carbon economy will fail. Therefore, in the period 2025–2029 they will price-in the resulting impact of increasing physical risks and extreme weather events on expected future output, earnings and dividend growth up to 2050. The pricing-in of these physical impacts is significant (up to 3% p.a.) and larger than in the Paris pathways.

However, in the Failed Transition Pathway, physical risks will increase very significantly in the second half of this century. At some point in the future, financial markets will start to price-in these very substantial post-2050 physical risks. We assume that this occurs in the second half of the 2030s. These impacts are not only larger (up to 8% p.a.) but also have substantial adaptation costs as well.

In the longer term, the Failed Transition Pathway is impacted by physical risks impacts as they occur. These become significant, rising to over 2% p.a. by 2060. Again, this limits the forecasting horizon of the model, as the physical damage now starts to overwhelm GDP projections and assumed relationships between the economy and markets are likely to break down. Furthermore, impacts on food production, water scarcity, migration and geopolitical risks (which are omitted from the modelling) are almost certain to be significant factors at or before this point.

Figure 9. Percentage difference in the median total equity return from 2020 to 2030 due to sectoral pricing-in shocks (relative to climate-uninformed baseline).