An independent catastrophe risk management and modelling firm has issued a report looking at the cost in insured losses that would be incurred today from some of history’s largest and most severe hurricane events. Karen Clark & Company (KCC) has published a report identifying historical US hurricanes that would likely cause $10 billion or more in insured losses were they to follow a similar track today.
Using their own proprietary methodology, KCC examined the nearly 180 hurricanes that have hit the U.S. since 1900 and ascertained that 28 of those hurricanes would result in $10 billion or more in insured losses in 2012 given the increased number, size and cost of structures in their paths.
The report is interesting as part of the catastrophe bond modelling process for hurricane risks usually involves modelling a cat bond against historic storms to see which would cause sufficient damage to trigger the bond. Using KCC’s methodology it appears that there may be many more storms that could cause impacts to cat bonds than other risk modellers methodologies might suggest.
Of course, all risk models are different and result in differing levels of impact and insured loss projections. You only have to look at the results of the remodelling of outstanding cat bonds with the new version of the RMS model to see how models can impact the expected loss of cat bonds.
The projected size of the insured loss from some of these historic hurricanes may come as a bit of a shock to many in the reinsurance market. Some are huge and if a storm ever has a similar impact it would be a reinsurance market changing event. Karen Clark & Co. provide details of some of the most impactful historic storms, in 2012 dollars using their methodology.
The largest potential insured loss would be a repeat of the 1926 Great Miami Hurricane which KCC estimate could cause a $125 billion loss if repeated today. The 1928 Okeechobee Hurricane and the famous Galveston storm of 1900, are the next costliest at $65 billion and $50 billion, respectively, both also enough to turn the market. Two other Florida storms, the 1947 Fort Lauderdale Hurricane and 1992’s Hurricane Andrew are also estimated at $50 billion. Rounding out the top loss producers are 1915’s Galveston ($40 billion), 2005’s Katrina ($40 billion), the 1938 Great New England ($35 billion), and 1954’s Hazel and 1965’s Betsy. both estimated at $20 billion. Hurricane Donna in 1960 affected the entire East Coast from Florida to Maine and would likely cause a $25 billion loss today. The remaining 17 storms on the list range from $10 to $15 billion each.
“It’s clear many hurricanes that struck the United States in the earlier part of the 20th century would cause orders of magnitude more damage today,” said Karen Clark, President and CEO, KCC. “This is due not only to an increased density of structures in coastal regions, but also to changes in construction practices that have resulted in larger and more expensive buildings. Even a storm as recent as Hurricane Andrew, which struck 20 years ago this month, would be three times as costly today.”
KCC claim that other sources of information on historical hurricane losses may underestimate the potential insured loss. They explain that other sources can adjust the original losses by the general rate of inflation to bring the numbers forward to current day values. This ignores the impact of increasing population, larger structures and the fact that building square foot construction costs tend to rise faster than the general rate of inflation. For example, simply adjusting the original Hurricane Andrew loss for inflation results in a current day estimate of only $22 billion (according to the Insurance Information Institute) versus $50 billion when these other factors are considered.
“We conducted this study because our clients asked us to provide historical event information as another tool they can use to assess their catastrophe loss potential,” said Glen Daraskevich, Senior Vice President, KCC. “Researching the various data sources we found incomplete and conflicting information, which led us to compile and thoroughly evaluate all of the sources. Our methodology built on previous work which we extended and enhanced to develop a credible and reliable estimate for each historical storm.”
KCC explain more about their methodology; Since each data source has limitations, developing robust estimates of the insured losses from historical hurricanes required multiple tools and data sources including knowledge of current property values, the size and extent of the hurricane windfields, and historical loss information. For this study, KCC combined approaches, gathering and cross-referencing information, modifying that information as necessary to ensure reliability, and using it to estimate the losses for each hurricane.
The output from this report indicates that the U.S. could face a $10 billion or greater insured loss from a hurricane every four years, a 25% probability. A $50 billion insured loss has a 5% probability according to the data.
“As this and our previous studies have shown, large hurricane losses are possible along the entire Gulf and East coasts,” added Ms. Clark. “Our clients can now estimate their losses for significant historical events as an additional perspective on catastrophe risk. This information is useful for many purposes including better understanding the magnitude of potential catastrophe losses in different regions, benchmarking the model output, and assessing the value of certain risk transfer options such as industry loss warranties (ILWs). Because catastrophe losses now dominate many of the property lines, the more credible scientific information that can be brought to bear on risk management decisions, the better.”
Of course we have not had a $10 billion hurricane insured loss for a number of years now, potentially meaning we are overdue. For reinsurers, issuers of catastrophe bonds and investors in hurricane ILS, this study drives home the need to understand your risks as well as you possibly can. If that means employing multiple catastrophe models then it would be recommended. To get a clear view of the risk for a hurricane cat bond it is prudent to understand the risks of attachment from historic storms and this study shows that it is worth ensuring you have considered every possible factor which could contribute to an increased probability of loss. The key thing to remember is that as long as the risk is well documented, well modelled and communicated to investors in a manner that they can understand, then there is no reason for a higher estimate of historic hurricane losses to prevent a cat bond issuance. It could of course affect pricing, premium or coupon paid to investors and comparable costs with traditional reinsurance, although the same data impacts reinsurance rates as well.