Article
The latest Intergovernmental Panel on Climate Change (IPCC) report, which was published in August 2021, had an entire chapter devoted to climate change and extreme weather events such as hurricanes, winter storms, floods, and droughts. The IPCC report summarizes and reflects the latest findings of the scientific and research community on the topic of climate change.
December 10, 2021
Niklaus Hilti, Head of Credit Suisse Insurance Linked Strategies,
Georges Bolli, Risk Aggregation & Management at Credit Suisse Insurance Linked Strategies
We took those findings that concern climate trends and tried to quantify their impact on the overall insurance industry, which includes the reinsurance and ILS markets. We believe that this report is the first of its kind to quantify the annual inflation of insurance losses due to climate change for the reinsurance and ILS markets. We also attempted to quantify the expected effects of a warmer climate that further increases extreme weather events on the (re)insurance industry over the next 20 years.
We believe that the annual inflation of losses for the (re)insurance industry due to climate change on property insurance lines is around 1.35% to 2.50%, depending on the exposure, region, type of natural peril, and seniority of the reinsurance transaction. The (re)insurance industry (including ILS) has not increased premiums sufficiently over the last two decades. In particular, the lower reinsurance premiums during the soft market phase between 2013 and 2017 were a heavy burden for margin adequacy. However, the challenging years from 2017 to 2021, with significant natural catastrophe losses, led to a steady increase in premiums on an almost global basis. This positive momentum was and still is necessary to adapt risk-adjusted premiums for reinsurers and ILS investors. Furthermore, the reinsurance market (including ILS) cannot afford to have lower risk-adjusted premiums going forward. Considering the inflation of insurance losses, risk-adjusted premiums will have to increase on average by at least 2% every year just to remain risk neutral (from a climate change perspective) in the future.
One of the questions asked most frequently by investors these days is whether risk models adequately reflect the climate change trend. We conducted an analysis of US hurricane risk, covering the period between 2006 and 2021. Our analysis showed that the inflation of insurance losses due to climate change is captured by the vendor model we included in our assessment.
Ultimately, we are convinced that it will be essential to apply strict measures and take decisive steps in managing the risks within ILS portfolios to make them resilient to inflation of insured losses caused by climate change. We believe that a combination of de-risking and higher premium levels is key for the reinsurance and ILS markets. Furthermore, the reinsurance and ILS markets have to react decisively now and demand annual increases in risk-adjusted premium levels in order to at least remain risk neutral with regard to climate change.
The IPCC is an independent body that comprehensively analyzes and summarizes current research on climate change. The IPCC’s Sixth Assessment Report, which was published in August 2021, included explicit assessments of extreme weather events around the world in a changing climate for the first time, including on a regional scale. In this latest report, the IPCC published its findings on the causal relationship between human-induced global warming and extreme weather events, such as droughts, extreme precipitation, tropical cyclones, and other storms. It is clear from the report that even relatively small incremental increases in global temperatures can result in disproportionate changes in extreme weather events.
In this report and the underlying analysis, we attempt to apply the latest findings in the IPCC report to the (re)insurance and ILS industry, to answer the following questions:
Chart 1 shows the exposure to different regions and perils in the cat bond and reinsurance markets. Perils in gray indicate exposures that are not weather related. Those in green indicate exposures that are weather related. Perils in blue indicate mixed exposures, which may be weather related. In our impact analysis regarding climate change trends, we included green and blue perils. Therefore, our analysis covers approximately 80% to 90% of the weather-exposed part of the cat bond market and approximately 70% to 80% of the weather-related or weather-exposed part of the property catastrophe reinsurance market.
1 Outstanding cat bond volume, excluding life and private cat bonds.
2 Estimated property natural catastrophe limits.
3 EQ = earthquake.
For illustrative purposes only. Data as of September 2021. Source: Credit Suisse
As indicated in Chart 1, the most relevant peril on a global scale (and thus for ILS and reinsurers) is represented by tropical cyclones (commonly known as hurricanes and typhoons), which aggregate the largest part of the reinsurance and ILS markets. Therefore, it is worth focusing on tropical cyclones in general, and US hurricanes in particular, to begin with.
Table 1 shows the relationship between hurricane wind speeds and their destructive power. The left-hand side of the table displays the Saffir–Simpson scale, which is a commonly used classification system for tropical cyclones. On the right-hand side of the table, the first two columns provide a comparison of the destructiveness of different hurricane categories compared to a Category 1 hurricane. The last two columns compare major hurricanes (Categories 3 to 5) versus an average of Category 1 and 2 hurricanes. The destructiveness of hurricanes, which is proportional to insured losses, is expressed by using the Hurricane Intensity Index (HII) or the Hurricane Hazard Index (HHI). As the comparison reveals, the destruction potential of major hurricanes is about 3.7 times the destruction potential of an average of Category 1 and 2 hurricanes.
| 1 minute max. sustained wind speeds | HII | HHI | HII (major to minor) | HHI (major to minor) | |||
| Category |
Miles per hour |
Kilometers per hour |
|||||
| Tropical depression | <39 | <63 | |||||
| Tropical storm | 39–73 | 63–118 | |||||
| Hurricane | Category 1 | 74–95 | 119–153 | 1 | 1 | 1 | 1 |
| Hurricane | Category 2 | 96–110 | 154–177 | 1.90 × | 2.64 × | ||
| Hurricane | Category 3 | 111–129 | 178–208 | 2.60 × | 4.30 × | 2.30 × | 3.70 × |
| Hurricane | Category 4 | 130–156 | 209–251 | 3.75 × | 7.30 × | ||
| Hurricane | Category 5 | >156 |
>251 | 5.10 × | 11.70 × | ||
For illustrative purposes only. Sources: Credit Suisse, IPCC, publication by Mehdi Rezapour and Tom E. Baldock from December 2014 in Weather and Forecasting, publication by Kossin et al. from 2020, modeling company
Consequently, increasing wind speeds have a disproportionate effect on insured losses. For major hurricanes, this can lead to a 130% to 270% increase in destructive force. Historically, major hurricanes caused more than 85% of the damage in the US, although they only accounted for 24% of landfalling hurricanes.
The IPCC report makes the following statement about tropical cyclones:
“It is likely that the global proportion of major (Category 3 to 5) tropical cyclone occurrences has increased over the last four decades, and the latitude where tropical cyclones in the western North Pacific reach their peak intensity has shifted northward; these changes cannot be explained by internal variability alone (medium confidence). There is low confidence in long-term (multi-decadal to centennial) trends in the frequency of all-category tropical cyclones. Event attribution studies and physical understanding indicate that human-induced climate change increases heavy precipitation associated with tropical cyclones (high confidence), but data limitations inhibit clear detection of past trends on the global scale.”
Key findings from the IPCC report:
In order to quantify the impact of tropical cyclones on the (re)insurance and ILS markets, we made the following assumptions from a climate trend perspective:
This assumption reflects the 6% trend mentioned above and aims to quantify the increased damage from hurricanes due to higher precipitation and a slower speed over land. There were several tropical cyclones over the last few decades that led to significant precipitation damage. Notable examples include Hurricane Harvey in 2017, Hurricane Florence in 2018, and Tropical Storm Allison in 2001. Additionally, in recent decades, increased rapid intensification of hurricanes was observed. Considering all these factors, we estimate the equivalent of an approximately 10% increase in major hurricanes, while leaving the total number of hurricanes constant.
Although a 10% increase over a decade does not indicate a strong trend, we tried to quantify its impact on the (re)insurance industry. To do so, we employed a vendor model commonly used in the insurance industry for calculating insured losses from hurricanes and modified it to make the incidence of major hurricanes 10% higher, while leaving the overall number of hurricanes constant in the simulation sample.
| Insurance industry loss triggers (in USD bn) | Increased default probabilities (including a 10% increase over ten years) |
|---|---|
| 20 | +14% |
| 40 | +17% |
| 60 | +18% |
| 80 | +19% |
| 100 | +23% |
| 120 | +24% |
| 200 | +25% |
| 250 | +26% |
For illustrative purposes only. Sources: Credit Suisse, vendor model
The simulated trend of a 10% increase in hurricane intensity per decade leads to increased default probabilities for the industry loss levels between USD 20 bn and USD 250 bn. The figures in Table 2 reveal an over-proportional effect in the tail (i.e. higher increase in probability of more severe catastrophic events).
As a result, breaking this impact down for the reinsurance and ILS markets, we calculated a relevant annual inflation due to this climate trend of 1.35% to 2.5%.
As a next step, we considered the following non-climate-related exposure growth trends:
The average non-climate-related exposure growth trend in the US over the last 30 years is estimated to have been between 3.4% and 3.9% per annum.
Therefore, we believe this climate trend, combined with non-climate-related growth trends, leads to an average increase in risks by 4.75% to 6.40% per annum for US hurricane risks. Consequently, reinsurers and ILS funds require on average a 4.75% to 6.40% increase in their risk-adjusted premiums each year in order to remain risk neutral.
Continuously increasing temperatures over the next few decades raise the concern that the inflation of insurance losses due to climate change could soon make up 60% to 75% of non-climate-related inflationary loss trends. Therefore, we believe that the overall inflationary loss trend (climate and non-climate related) could lead to an annual increase of 5.5% to 7.5%, which would need to be passed on to insurers and, ultimately, to consumers.
We conducted a comparison of risk modeling output between 2006 and 2021 using a cat bond issued in 2006. Table 3 shows the risk figures from 2006 and the risk figures based on today’s model and the respective delta.
| 2006 model analysis (annual expected loss) | 2021 model analysis (annual expected loss) | Delta 2021/ 2006 model inflation (annual) | Delta 2021/ 2006 model inflation (annual) | Estimated climate- and non-climate-related inflation (annual) | |||
| Long term | Warm SST4 | Long term | Warm SST | Long term | Warm SST | ||
| Tranche 1 | 1.79% | 2.69% | 4.30% | 4.90% | 6.02% | 4.08% | 4.75%–6.40% |
| Tranche 2 | 2.28% | 2.97% | 5.30% | 6.00% | 5.78% | 4.80% | 4.75%–6.40% |
| Tranche 3 | 3.02% | 3.79% | 6.80% | 7.70% | 5.56% | 4.15% | 4.75%–6.40% |
| Tranche 4 | 3.97% | 4.96% | 9.10% | 10.40% | 5.69% | 5.06% | 4.75%–6.40% |
4 SST = sea surface temperature.
For illustrative purposes only. Data as of September 2021. Sources: Credit Suisse, modeling company, cat bond offering document
The comparison indicates an annualized increase in modeled risk of around 4% to 6% over the last 15 years. Therefore, the risk model included in our analyses is adjusted for inflation of construction, inflation of contents and wealth, and population growth, and it seems that the model also reflects the inflation of losses due to the current climate trends.
Acknowledging that the risk model captured the overall inflationary loss trend (including climate- and non-climate-related trends) over the last 15 years, the obvious next question would be whether reinsurers and ILS managers were able to reflect the modeled loss increases in adequate premium increases.
To answer this question, we tried to compare transparent, like-for-like structures.
| 2006 model analysis (annual expected loss) | Estimated coupon 2006 | 2021 model analysis (annual expected loss) | Estimated coupon 2021 | |||
| Long term | Warm SST | Long term | Warm SST | |||
| Tranche 1 | 1.79% | 2.69% | 10.0% | 4.3% | 4.9% | 10.0% |
| Tranche 2 | 2.28% | 2.97% | ≈12.5% | 5.3% | 6.0% | 11.0% |
| Tranche 3 | 3.02% | 3.79% | ≈14.5% | 6.8% | 7.7% | 12.5% |
| Tranche 4 | 3.97% | 4.96% | ≈16.0% | 9.1% | 10.4% | 14.0% |
For illustrative purposes only. Data as of September 2021. Sources: Credit Suisse, modeling company, cat bond offering document
In Table 4, it appears that over the past 15 years, reinsurers and ILS (cat bond) managers were not able to charge additional premiums despite increasing risks. As a result, inflation and climate trends came at a cost. This means that the increased risks reduced the margin of ILS investors.
Similarly, in Chart 2, we can see that the premium levels in the industry loss warranty (ILW) market also failed to increase in line with the increasing risks (from climate- and non-climate-related inflation).
Historical ILW rate for US all natural perils at a USD 50 bn industry loss trigger
For illustrative purposes only. Data as of 30.09.2021. Sources: Credit Suisse, broker prices, Guy Carpenter, Aon Benfield, Willis Re
The premium development for an ILW at a USD 50 bn industry loss trigger in the US due to a natural catastrophe (including hurricanes) shows that the premiums today are comparable to those eight years ago (in 2013), while the risks have steadily increased, as indicated by the expected loss trend line. Applying those increased trend risks, the risk-adjusted premium would need to be significantly higher today.
Insurance is slightly less affected by climate-change trends in terms of hurricanes as insurers have net exposures, which are less in the tail. However, they transfer more risk to reinsurers and ILS investors, meaning they will ultimately face the increased costs of reinsurance and risk transfer to ILS. Insurers will likely see increased costs with a five- to ten-year delay. In many states, insurance premium increases and deductibles are regulated, and as such, insurers are not free to charge policyholders for risks inflated by climate-change trends. According to our high-level analysis, in the state of Florida over the last 15 years, insurers were not permitted to make any significant increases in deductibles or rate changes to cover such inflationary trends. Therefore, we believe the risk-adjusted pricing of insurance policies is at least questionable.
Reinsurers seem to be overlooking these trend risks, as deductibles have not changed significantly and rate increases are not sufficient to keep up with climate- and non-climate-related inflation, meaning risk-adjusted margins have actually decreased.
ILS and cat bond market: In the overall ILS market, margins have been decreasing in a similar vein to those of reinsurers. However, deductibles in the cat bond market have been moving up, resulting in lower risks within cat bonds. As such, the cat bond market is in a favorable position (and is also on the cusp of securing further margins). The key is that over the next few years, risk-adjusted premiums will need to increase by at least 2% to keep up with loss inflation. Nevertheless, the signs are still positive, offering a way to keep ILS investments attractive.
Looking at the chart below, which shows the outstanding cat bond market as presented by Lane Financial, we can see the modeled expected loss in dark blue and the moving average of the yield spread represented by the dotted line. The fact that expected loss is currently only 3% – despite increased modeled losses having more than doubled over the last 15 years – demonstrates that even if investors are getting a smaller margin today, the market has seen significant de-risking, meaning there is still an attractive margin. However, we believe that over the coming decades, premium increases and/or de-risking will be pivotal in keeping up with climate- and non-climate-related inflation.
Historical cat bond yield spread and expected loss (EL)
Data as of 30.09.2021. Source: Lane Financial
The IPCC report shows that the expected frequency of heavy precipitation over land (30% higher compared to preindustrial levels) and agricultural droughts (70% higher compared to preindustrial levels) have already increased significantly. Depending on future global warming levels, these trends will intensify further, and to a substantial degree. Consequently, the risk of floods and wildfires is heavily linked to climate trends. We would like to draw attention to the fact that for both risk categories, we are not merely observing a trend related to severity distribution, such as the one for tropical cyclones discussed above; instead, we can see trend risk increasing, in terms of frequency and severity.
When modeling the potential impact on insured losses, we factored in increased loss distribution from the increases in frequency and severity. Therefore, we assumed that the increased frequency of heavy precipitation is proportional to the increased risk of floods. We made the assumption that the increased frequency of agricultural droughts is proportional to the increased risk of wildfires. Given that the statements in the IPCC report concerning agricultural drought are only for regions that are facing an increased drought risk, our calculations are just approximations, with the aim of understanding the effects in the tail, which is relevant for reinsurers and ILS funds.
| Original risk from 1850 to 1900 | Increased risk 97th percentile = 1 in 33 years | Increased risk 99th percentile = 1 in 100 years | |||
|---|---|---|---|---|---|
| Global temp. | +1°C today | +1.5°C est. 2040 | +1°C today | +1.5°C est. 2040 | |
| Flood risk | 1 | +32% | +54% | +37% | +57% |
| Wildfire risk | 1 | +82% | +124% | +84% | +129% |
Source: IPCC report
An interesting aspect of the above table is that if we look at the increased risk of a temperature increase by 1°C compared to preindustrial levels, we could argue that we are only looking at a risk increase of 32% to 84% over a period of more than 120 years. As such, this trend could be seen as negligible. However, we need to take into account that this trend was not linear in the past. Most of the effects were only observed in the last 30 to 40 years. More importantly, we need to be aware that we are most likely going to reach an increase in temperature of 1.5°C by 2040, regardless of the measures adopted to reduce greenhouse gas emissions. Consequently, we need to consider a further increase of risk by 20% during this period.
With reinsurance transactions attaching at lower levels (e.g. where industry losses in the US are attaching at levels of up to USD 15 bn), reinsurers and ILS managers are exposed to all perils, including hurricanes, wildfires, floods, severe convective storms, and winter storms. Each of these perils is likely to experience an inflation of losses as a result of the climate trend, in addition to other inflationary trends that are not climate related. Therefore, the loss inflation of such transactions sums up loss inflation for the different perils. Hence, reinsurance transactions at such low levels will be a no-go area going forward, and even transactions attaching at higher levels have to be monitored carefully for adequate rate increases to stay risk neutral. Each region (e.g. Europe, Japan, Australia, etc.) has a different definition of what has to be considered “low level.”
The risks associated with climate-related trends can be reduced to a minimum by investing in transactions covering single perils only (e.g. tropical cyclones). Any additional peril in a transaction may lead to an over-proportional increase of climate-related trend risk. Hence, we are in favor of clean structures with limited perils covered going forward to be able to manage associated trend risks accordingly. Many cat bonds only cover peak perils (e.g. earthquakes and/or tropical cyclones) and therefore offer a cleaner structure compared to a traditional reinsurance transaction covering “all natural perils,” which could include secondary perils (e.g. wildfires) that are also significantly affected by climate change.
In general, reinsurance transactions with aggregate cover are increasing the complexity of the different loss trends. We currently assume that if a reinsurance transaction with occurrence cover would have to see a 4.75% to 6.40% premium increase on a year-by-year basis, reinsurance transactions with aggregate cover would probably require an annual premium increase of 6% to 9% just to stay risk neutral. As such, we are in favor of occurrence transactions going forward to limit complexity in our portfolios.
As overall risk assumptions in the natural catastrophe (re)insurance business are expected to change, and with nonstationary climate risks evolving and adding more complexity, the importance of maintaining a sophisticated understanding of these trends and translating those into progressive underwriting capabilities is becoming more and more important. In our capacity as an ILS investment manager, we believe that we are in a good position to take proactive measures to respond to these trends and sustainably manage ILS portfolios for the challenges ahead.
Important information
Please note that our analysis/assessment was calculated based on one of the most widely accepted risk models for assessing natural catastrophe risks. Therefore, the resulting conclusion refers only to this model and cannot be applied to other risk models used in the (re)insurance industry.