Insurance looks to climate scenarios, but are they the right tool?

When journalists and policymakers talk about the risk climate change poses to insurance markets, they often focus on the threat to consumers: the fear is increasing and unpredictable severe weather events, perhaps caused by climate change, will make coverage either too expensive or simply unavailable for businesses and households.

But severe weather can sink insurers, too. In the US, Louisiana saw four hurricanes in 2020 and 2021 – as a result, a dozen insurers in the state went bankrupt between July 2021 and February 2023. Florida has also seen a slew of carriers fail in recent years, with a state resolution authority listing 14 property/casualty (P&C) failures since the start of 2020.

Last month the National Association of Insurance Commissioners (NAIC) published its National Climate Resilience Strategy for Insurance. Solvency testing was one of its five priorities, with a focus on using “scenario analysis”, a form of modelling, to quantify and guard against that risk.

Climate scenario analysis is not new – ironically, fossil fuel companies have a long history of trying to project how climate change will affect their infrastructure and assets. However, it is not clear whether present climate scenario tools are useful for measuring solvency – they involve different data, assumptions and timeframes. It also remains unclear exactly whether and how climate change shapes individual weather events, separate from other systems like El Niño.

Creating a clear link between solvency, weather and climate is a complex exercise dependent on hypotheses and projections – and so, inescapably, is a work in progress.

Capital at risk?

Ilana Winterstein, an environmental campaigner with the Sunrise Project, a global climate justice campaign network, says climate change presents a serious threat to the solvency of insurers because it “hits both strands” of their business.

“As underwriters they risk huge natural catastrophe payouts due to climate damage and as investors in the fossil fuel industry they risk stranded assets as environmental regulations will increasingly restrict oil and gas production,” she says.

In a survey of 44 European non-life insurer enterprises, published in 2022, the European Insurance and Occupational Pensions Authority (Eiopa) found these firms had €42.6trn ($45.5trn) in windstorm cover, as well as €38trn in river and coastal flood cover and €22.8trn in wildfire policies.

Norbert Pieper, a spokesperson for Germany’s federal financial regulator, the Bundesanstalt für Finanzdienstleistungsaufsicht (BaFin), says P&C insurers are mainly exposed to physical risks, while transition risks are more of a problem for life insurers.

“There is a big challenge with climate change… [the Network for Greening the Financial System and Intergovernmental Panel on Climate Change scenarios] just provide climate assumptions on global warming, but you can’t really directly link those to what will happen to different periods”
_{Charles-Marie Delpuech
S&P Global Ratings}

Other regulators, including the central banks of South Africa and New Zealand and Japan's financial regulator, tell Insurance Day they interpret climate change as a risk to insurer solvency. Italy’s insurance regulator, the Istituto per la Vigilanza sulle Assicurazioni (Ivass), says it is “aware of the risks” posed by climate change to supervised insurers’ solvency and has been “constantly adapting” to EU regulation to encompass these new risks within its prudential supervisory tasks.

However, although severe weather and climate change can contribute to solvency risks, analysts argue in most cases it will not alone capsize a competently run insurer, especially not a larger one. Robert Chaplin, partner at law firm Skadden, Arps, Slate, Meagher & Flom, says climate change could cause “significant financial strain” to carriers in vulnerable zones, but a well-managed insurer is unlikely to collapse from its impact alone.

“Climate change in itself is not a risk to insurers’ solvency – the risk arises from how an insurer manages the effects of climate change, both with respect to physical climate risk and carbon transition risk,” Brandan Holmes, vice-president and senior credit officer at rating agency Moody’s, says.

Taos Fudji and Charles-Marie Delpuech, directors for Europe, the Middle East and Africa insurance ratings at S&P Global Ratings, tell Insurance Day they agree climate change-induced severe weather events may prove fatal for small property carriers in high-risk areas. However, they do not pose immediate threats to large, diversified insurers, which can also adjust their underwriting exposure every year or two.

Regulators’ warnings

Many insurance regulators are now pressing companies to assess climate change risks as part of their corporate governance. In September 2020, New York state’s insurance regulator, the Department of Financial Services (DFS), instructed insurers to start considering climate risks.

In April 2021, Eiopa told national insurance regulators they “should expect insurers to integrate climate change risks in their system of governance, risk management system and Orsa [Own Risk and Solvency Assessment]”. BaFin, the Dutch central bank and Ivass all told Insurance Day they expect insurers to analyse climate risks in their Orsa exercise.

In November 2021, the Australian Prudential Regulation Authority (Apra) published a “prudential practice guide” informing the firms it regulates the agency “considers it prudent practice for the board to seek to understand and regularly assess the financial risks arising from climate change that affect the institution, now and into the future”. Apra is currently conducting a climate risk assessment survey, its second, across the financial services sector.

In Quebec, Canada the province’s financial markets regulator, the Autorité des Marchés Financiers (AMF), says the firms it regulates “are now expected to consider climate-related risks in their integrated risk management processes”.

Estimating catastrophe

To measure the risk from global warming, regulators and other experts recommend or require the use of scenario analysis. The International Association of Insurance Supervisors (IAIS) defines scenario analysis as “a method of assessment that considers the impact of a combination of circumstances to reflect historical or other scenarios, which are analysed in light of current conditions”. The IAIS held a consultation on the use of scenario analysis in gauging climate risks to insurers between November 2023 and February 2024.

In the US, the NAIC strategy calls on state regulators to “embed climate stress testing and climate scenario analysis into routine financial analysis, data collection and financial surveillance”.

Some US state regulators now require insurers to conduct climate scenario analyses. In guidance issued in November 2021, New York’s DFS said it would expect insurers to “use scenario analysis to inform business strategies and risk assessment and identification. Scenarios should consider physical and transition risks, multiple carbon emissions and temperature pathways and short-, medium- and long-term horizons”. Neighbouring Connecticut issued an identical guideline in September 2022.

In the EU, Eiopa guidance from August 2022 expects regulators to require firms “to consider at least two long-term climate scenarios”: one where temperatures rise no more than 2°C and one where they rise more than that.

In guidance issued last year, the Reserve Bank of New Zealand advised the firms it regulates, including insurers, to “develop capabilities in climate-related scenario analysis and stress testing or have access to external scenario analysis and stress testing capabilities while internal capacities are developed”.

The central bank tells Insurance Day it encourages but does not require insurers to use scenario analysis. However, New Zealand law requires larger insurers to show “how they have analysed the resilience of their business model and strategy under a 1.5°C scenario, a more than 3°C scenario and another scenario of an entity’s choosing, which will assist their management of solvency risks”.

Japan’s Financial Services Agency (FSA) and the Bank of Japan piloted a climate scenario analysis with three banks and three non-life insurance companies, building on scenarios from the Network for Greening the Financial System (NGFS), an international association of central banks and financial regulators, in 2022. Japanese regulators concluded global warming would increase claims and are working on a second scenario analysis, but the FSA tells Insurance Day it does not currently require insurers to use scenario analysis as work on methodology and data collection is still ongoing. However, "we consider it desirable for financial institutions to eventually utilise scenario analysis", and notes some non-life insurers in Japan have done so.

The AMF tells Insurance Day it expects insurers to start using climate scenario analyses from the second half of 2024. It will also publish a climate risk management guideline this year.

Writing the scenario script

Regulators and insurers can write their own scenarios, but frequently rely on scripts produced by external associations. One common source is the NGFS.

The NGFS has developed seven scenarios grouped into four categories:

• Orderly, which “assumes climate policies are introduced early and become gradually more stringent”;

• Disorderly, which expects policies are adopted more slowly and with less co-ordination;

• Hot house world, in which too little is done and “critical temperature thresholds are exceeded”, and

• Too little, too late, which also involves tardy and unco-ordinated policies and thus “elevated transition risks in some countries and high physical risks in all countries due to the overall ineffectiveness of the transition”.

The scenarios run to 2060, and project temperature changes ranging from 1.4°C – for example, if net zero is reached by 2050 – to 3°C or more if the world maintains current trends.

The Intergovernmental Panel on Climate Change (IPCC) listed five scenarios in its 2021 report, with escalating degrees of global heating – from 1.4°C to 4.4°C by 2100.

In its 2021 Biennial Exploratory Scenario, the Bank of England proposed three scenarios – early action, late action and no additional action – which adopted 2050 as an end date. In the first two scenarios, temperatures rise no more than 1.8°C over this period; in the last, temperatures rise 3.3°C.

The Dutch central bank has published guidance for internal scenario analyses. It recommended firms employ “the principles of relevant stress tests as a basis for assumptions” in their scenarios, citing exercises conducted by Eiopa, the Dutch central bank and the UK’s Prudential Regulation Authority (PRA).

Ivass tells Insurance Day Italian carriers usually pick two NGFS scenarios, “one of orderly transition and the other of disorderly transition to assess the materiality of their exposure to both physical and transition risks”.

“Sometimes, insurers adapt these scenarios depending on the characteristics of their portfolios,” Ivass adds.

Last month the PRA published a paper advising insurers how to adapt macro-level scenarios to individual asset types.

Canada’s federal Office of the Superintendent for Financial Institutions and the AMF are jointly developing a Standardised Climate Scenario Exercise (SCSE), with the second part of a consultation beginning last month. The SCSE includes three sections dealing climate transition risks, covering credit, market and real estate exposure risks, as well as a physical risk section. The AMF says the SCSE “is not an exercise to obtain an absolute quantitative impact of climate change or predict the future, but to differentiate among risks”.

But is this the right tool?

Scenario analysis can help map risks to assets and property over time, but John Scott, head of sustainability risk at insurance giant Zurich, suggests climate scenario analysis is not a useful tool for assessing solvency risk.

One problem is differentiating between weather and climate. The existence of climate change is “incontrovertible”, Scott says, but the relationship between global warming and individual weather phenomena “is not as straightforward as we think”, he tells Insurance Day. “These are complicated issues of attribution and climate science, which are difficult to interpret,” Scott continues.

Delpuech says it is difficult to measure exactly how much climate change contributes to rising losses from extreme weather, partly because the precise impact of climate change on the frequency and severity of events is not well quantified. Fudji, referring to Swiss Re data, says in recent years natural catastrophe losses have risen. “The very high majority – I would say more than 80% – of the increase in the sum of insured losses for insurers is linked to the rise in property values,” he says, as opposed to more frequent and more severe extreme weather events.

Another problem Scott identifies is climate scenario analysis uses very different timescales and data from solvency assessments. Regulators and firms measure their solvency on an annual basis or perhaps over a few years, while climate scenarios run for decades. And while insurers measure things like credit or liquidity risk based on retrospective data, such as claims records, climate scenarios rely on hypotheses and projections, with no guarantee the risks will ever be realised.

“There is a big challenge with climate change,” Delpuech says. The NGFS and IPCC provide climate scenarios, “but the challenge is those scenarios… just provide climate assumptions on global warming, but you can’t really directly link those to what will happen to different periods” along the scenario timescale.

Delpuech adds IPCC and NGFS scenarios “don’t give any indication of how the physical risks linked to global warming actually play out… there is no quantification”. Insurers and regulators must develop more exact specifications to gauge their solvency.

Another issue is whether analyses of long-term climate change risks should change the ways insurers allocate capital. Solvency management involves maintaining sufficient capital in today’s insurance markets to protect policyholders and for financial stability. That works very well for the existing severe weather-related risks, “but it could be destabilising to current pricing and markets, if regulators require capital to be applied today to something that may or may not happen” decades down the line, Scott says.  

Holmes says “overestimating risk could lead to insurers forgoing profitable business and become uncompetitive in the market”.

“The more immediate risk of greater unpredictability of severe weather can be managed through more sophisticated data and modelling, which helps price risk more accurately,” Holmes adds.

Fudji says insurers can reprice or drop policies or acquire more reinsurance on a regular short-term basis – unlike banks, which have fixed-term assets in the form of loans. It is “not necessarily most relevant to make business decisions today on what could happen in 25 or 30 years’ time”, he says.

Scenario analyses may be plausible forecasts, but one cannot determine how probable they are, Scott continues. If insurers start reserving capital for hypothetical long-tail climate risks, they will end up severely curtailing the availability of affordable insurance.

Scott distinguished between solvency requirements, which tend to rely on historical data, and scenario analysis, which is more prospective and lends itself to dealing with uncertain pathways and testing hypotheses. He suggested that regulators apply the latter instrument, as encouraged by the Taskforce on Climate-related Financial Disclosures with its long-term “what-if” guidelines. “I think everybody’s learning still,” he said, “and especially in the regulatory space.”