Hurricane frequency in Florida and the potential impact of climate change

Hurricane frequency in Florida and the potential impact of climate change

With global warming receiving increased media attention, the record insurance losses caused by natural catastrophes in 2017 and 2018 have made the general public, and the insurance community in particular, question a potential relationship between climate change and increasing insurance losses. 

And, in fact, it is worth examining how climate- and human-related factors influence insurance losses and how future expected trends relating to climate change may impact the insurance sector. To that end, we will also look at the challenges that insurers and reinsurers face and the potential implications this may have on the insurability of property damage due to natural catastrophes.

Global climate change and natural catastrophes

Historical insurance losses from large-scale extreme weather-related events demonstrate that many of the most costly events have occurred in recent years, which would indicate that there is a trend towards recent catastrophe losses being significantly higher than past losses. This is exemplified by the fact that the insurance losses from 2017 and 2018 – more than USD 180 bn combined – were both higher than ever before. Strong tropical storms occurred in both years, mainly in the US and Japan, and caused severe damages. Going from bad to worse, after the storm seasons had ended in both the Atlantic and the Pacific, wildfires in California inflicted severe destruction with insurance losses far beyond historical levels in both years.

Despite the strong foundation of scientific evidence that global warming is also linked to the human-caused increase in greenhouse gases, there is still considerable uncertainty regarding its impact on severe-weather-related events such as hurricanes, storms and floods. In fact, historical data on hurricane frequency and landfall frequency, for example, illustrate that there is no discernible trend of increasing hurricane frequency or greater landfall frequency today. However, our global climate has undergone long-term cycles long before human-induced increases in greenhouse gases in the atmosphere. In the North Atlantic, for instance, where most of the hurricanes that make landfall in the US originate, sea surface temperatures are subject to a climate cycle called the Atlantic Multidecadal Oscillation (AMO). This cycle is related to ocean currents in the North Atlantic and has a direct link to sea surface temperatures. The AMO is characterized by warm and cold phases that usually last between 20 and 40 years – i.e. several decades. Since the mid-1990s, the AMO has been in a warm phase, meaning average sea surface temperatures have typically been higher than the long-term average. Historically, such “warm phases” have often resulted in a higher frequency of hurricanes in the North Atlantic because storms get stronger as they move over warm waters.

When we compare the development of the AMO over the past decades with hurricane activity in the North Atlantic (made up of the number of hurricanes per year, including those that have not made landfall), we can observe a comparable pattern. These climate cycles make it very difficult for scientists to attribute the changes in hurricane frequency to either natural climate cycles or human-induced global warming.

Atlantic Multidecadal Oscillation (AMO) compared to the number of hurricanes

The graph shows the Atlantic Multidecadal Oscillation (AMO) compared to the number of hurricanes per year from 1856 to 2016. Since the mid-1990s, the AMO has been in a warm phase, meaning average sea surface temperatures have typically been higher than the long-term average. In line with these developments, the number of hurricanes per year has also increased since the mid-1990s.

Sources: National Oceanic and Atmospheric Administration (NOAA), Credit Suisse

Looking at hurricane frequency alone shows that natural climate cycles are responsible for longer-term variability in hurricane occurrence rather than climate change. However, with respect to North Atlantic hurricanes, scientific research suggests that there has indeed been a slight increase in the frequency of very severe hurricanes (i.e. those of Category 4 and 5 on the Saffir-Simpson hurricane wind scale) since the 1970s.

This trend in increased Atlantic hurricane intensity is likely related to higher sea surface temperatures in the North Atlantic observed over the past decades. Although this is also partly driven by the AMO, climate change has been proven to generally increase sea surface temperatures. Typically, hurricanes strengthen more and faster when moving over very warm sea waters – the warmer the sea surface temperature, the stronger the hurricane can become. This not only results in higher wind forces from the hurricane, but also in exponentially higher water evaporation rates due to higher atmospheric humidity, which fuels hurricane intensity and leads to more severe and prolonged rainfall from hurricanes after landfall. Recent examples include Hurricane Harvey in 2017 and Hurricane Florence in 2018, both of which caused severe flood damage due to intense rainfall.

In summary, it is likely that climate change impacts hurricane activity – they may not become more frequent, but we have observed a slight increase in very severe hurricanes, which can be partly attributed to higher sea surface temperatures driven by global warming. A direct relationship between global warming and insurance losses caused by hurricanes, however, is more challenging to prove.

The human factor

Over the past decades, the total value of insured properties at risk from natural disasters has been strongly influenced by demographic factors such as growing wealth levels and increased population densities in hazard-prone areas such as the east and southeast coasts of the US.

US Population by Region

The chart shows US population growth in six regions from 1940 to 2016. The majority of the population growth is in coastal areas. For example, the population of Florida, the US state most at risk from Atlantic hurricanes, has grown from 2.8 million in 1950 to 21.3 million in 2018 – a 660% increase. The sharpest population increases are reflected in the Atlantic, Gulf Coast and Coastal West.

Source: Colorado State University, Phil Klotzbach

The population of Florida, the US state most at risk from Atlantic hurricanes, has grown from 2.8 million in 1950 to 21.3 million in 2018 – an increase of 660%, with the majority of the population now living in coastal areas. This combined with an increase in wealth levels and increased insurance penetration, results in significantly higher “at risk exposure” today compared to, for example, the 1950s.

To illustrate this significant impact of changing exposures on insurances losses, the Credit Suisse ILS team did an analysis more than 10 years ago based on Hurricane Hazel. This Category 4 hurricane, which made landfall at the border between North and South Carolina in 1954, caused economic losses of USD 281 mn at the time. Under today’s conditions, this would be considered a minor event. However, if a hurricane like Hazel struck again in 2008 (i.e. 54 years later), the losses would be as high as USD 31.5 bn (considering the impact of inflation, rise in real wealth, and increase in housing density as of 2008).

Adjusted economic losses for Hurricane Hazel

This chart shows the adjusted economic losses from Hurricane Hazel. At the time of occurrence in 1954, the storm caused economic losses of USD 281 mn. In 2008, 54 years later, the same event would cause more than USD 30 bn in economic losses. The main drivers of the increasing losses are inflation, followed by the increasing housing density and rise in real wealth.

Source: Credit Suisse

The impact of global warming on insurance

With global warming expected to continue this century, there is a likely scenario of a further increase in the severity of North Atlantic hurricanes, which can result in greater damages when these storms make landfall on the US coast, especially due to the higher flood/rainfall risk. Insurance companies are therefore faced with the challenge of dealing with greater uncertainty in terms of risk assessment and potential constraints in their insurance capacity. This may impact the insurability – the availability and affordability of insurance in the private sector – of certain properties, which is determined by several factors.

Firstly, if natural catastrophes are insurable events, one key aspect is the ability of insurance companies to properly assess and quantify loss amounts and probabilities. With expected future trends in changing weather patterns, catastrophe risk assessment will likely become more reliant on modeled predictions to supplement the shortcomings of relying on past experiences.

Secondly, the estimated loss probabilities will determine whether insurance can be covered by sufficient capital and at affordable premium levels in the market. A higher perceived insurance risk could lead to higher premiums and/or more restrictive underwriting for certain high-risk properties.

For reinsurance and ILS, this means that all of the risks involved with climate change will not necessarily be accepted by the reinsurance market. More restrictive underwriting could lead to properties becoming uninsurable in the private market, leaving risks uninsured or insurance coverage shifting to the public sector instead. One example of this is flood insurance in the US: while insurance companies provide homeowners with coverage against property damage from hurricanes, this coverage often excludes flood-related damages e.g. from storm surges or flooding caused by heavy rainfall. However, homeowners can buy additional “flood coverage” from the National Flood Insurance Program (NFIP), a state-sponsored insurer. As an example: Hurricane Harvey, which struck Texas as a Category 4 hurricane in 2017, caused approx. USD 19.4 bn in property damage that was covered by the private reinsurance market, and USD 8.9 bn in flood-related damages covered by the NFIP.

In conclusion, in our view, the risk of slow-moving and very long-term trends such as global warming is limited since reinsurance contracts underlying ILS transactions typically have a maturity of 12 months. Thus, the latest scientific findings, which also feed into the risk models, can be reassessed on an annual basis during insurance trade renewals.

“Science versus perception”

While scientific evidence and historical data on hurricane landfall and frequency may be strong and clearly show that there is no trend of increased hurricane activity, people’s perception is very different – which is understandable given the events of 2017, 2018 and 2019.

In late August 2019, Hurricane Dorian was expected to make landfall in Florida as a major hurricane of Category 3 or higher. This would have made Dorian the third major hurricane in as many years to make landfall in Florida – and, of course, with correspondingly severe damages and significant insurance losses. Although Hurricane Dorian turned northwest and did not make landfall in Florida, it is understandable when people assume that we will see another Hurricane Irma, Michael or Dorian year after year. As the saying goes, “perception is reality.” Media coverage of hurricanes today begins many days before landfall, with frequent updates and sometimes 24/7 coverage of events.

But even when we turn to historical data, it becomes clear that we have just experienced one of a few rare clusters of very severe hurricane landfalls in Florida. Previously, between 1917 and 1919 and between 1926 and 1928, Florida twice experienced two landfalls from very severe hurricanes within a span just three years; from 1945 to 1950, the state experienced five landfalls from very severe hurricanes within just six years. And now it has gone through two landfalls in a row in 2017 and 2018 – and a “near miss” in 2019. However, the period leading up 2017 was remarkably inactive along Florida’s coast, with a consecutive 11 years without hurricane landfall – the longest such period on record. But for most people, the last few years are more deeply imprinted on their minds than the 11 years prior.

Hurricane landfalls in Florida since 1851

Since record-keeping began, Florida has experienced 13 very severe hurricanes measuring Category 4 or 5 at landfall.  This chart shows the hurricane landfalls in Florida from when record-keeping began in 1851 until 2018. The hurricanes are separated between categories 1–3 and 4–5. It illustrates the fact that, since record-keeping began, Florida has experienced 13 very severe hurricanes measuring Category 4 or 5 at landfall. The main point of this chart is to display perception versus reality in the sense that people today assume that the hurricane frequency has increased significantly and that they can expect years like 2017 and 2018 to repeat every year; however long term statistics do not show a trend. In fact, before 2017, hurricane activity was remarkably low along Florida’s coast, with 11 consecutive years of no landfall – the longest such period on record.

Source: North Carolina Climate Office, Credit Suisse; data from 1851 to 2018