As we move deeper into another Atlantic hurricane season, many in the insurance and reinsurance industry find themselves asking if this year will see an end to the very benign period of losses we have witnessed over the past decade. Since 2005, no major hurricane (Category 3 or higher) has made landfall in the U.S.—the longest stretch in recorded history. Overall, on a global basis, catastrophe losses have been far below average during the last decade, notably contributing to the very challenging market conditions in which many companies are operating today.
The risk, of course, is still there. However, our memories are short and, as I discussed in my previous blog, a sense of complacency has set in since the game-changing hurricane seasons of 2004 and 2005. Other than a major event actually occurring, what can shake us out of our false sense of security?
The framing effect
One imperative is that we view the risk through the proper lenses. To that end, I would recommend that everyone in the industry go beyond communicating catastrophe losses in terms of annual probabilities and return periods, and start speaking of losses over time horizons relevant to their businesses.
I am sure no company operates with a one-year planning horizon, yet in most discussions about catastrophe risk clients do exactly that. When considering the "1-in-100 return period loss" or the "loss at the 1% exceedance probability," only the likelihood of that size loss or higher occurring in the next year is being assessed.
When model results are expressed in this way, there is a reflexive tendency to discount such losses. The probabilities seem so low that some can incorrectly assume such losses will not occur in our careers or over a reasonable time horizon. In fact, it is a common cognitive bias termed the "framing effect," and we all need to be on guard against it.
Let's look at this in the context of annual industry insured losses for hurricanes impacting the U.S., keeping in mind that AIR estimates the industry aggregate loss at the 1% exceedance probability to be approximately USD 150 billion. If I offered you the chance to sell one of these two "contracts," each with the same premium and each lasting for 10 years, which would you pick?
Contract A has a 1% annual probability of a loss equal to or greater than USD 150 billion
Contract B has a nearly 10% chance of a USD 150 billion loss or greater in the next 10 years
Many people would pick Contract A, but the risk is, of course, equivalent. Now do the same exercise for the risk in your own portfolio or a contract. You should be indifferent to whether the risk is presented on an annual basis or over some time horizon.
If, however, you are increasingly anxious when the risk is presented over longer time horizons, you may have fallen into a framing trap and are likely underestimating your tolerance to that risk, or you may not have devoted enough attention to how much you are willing to lose in that time period. Of course, the time horizon you pick should be meaningful to your business—there is nothing magical about 10 years in this example.
Allowing for growth
If you do start to express your catastrophe model results over certain time horizons and you expect to maintain market share, you should take it one step further and factor expected growth in to the number and value of the exposures over that time period. To put this in perspective, AIR expects the increase in the total insured exposure in the U.S. (meaning the increase in both the number and value of insured properties) will average almost 6% per year over the next decade.
One implication of this growth is that the same hurricane today that would cause roughly USD 25 billion in loss will cause more than USD 40 billion in loss 10 years from now. Even more striking is that the probability of a year with USD 150 billion or more in insured hurricane losses over the next decade goes from roughly 10% to nearly 17%. That kind of statistic is likely an eye-opener for many.
Many companies will be reviewing their catastrophe model results in the coming weeks and months. If you aren't doing it already, I suggest you use that next review as an opportunity to present the losses both in terms of annual probabilities and over a time horizon relevant to your business, with expected exposure growth factored in.
If we can start doing this as common practice among industry professionals, it opens the door to communicating in this manner to a much broader audience—your insureds. Individual property owners are especially prone to the "framing effect." We'll be doing them a service by framing the risk in a way that is more meaningful to them.
Catastrophe models are excellent tools for helping to understand risk in a deeper and more meaningful way. However, just as we can get used to experiencing low hurricane losses in the U.S., we can also get used to looking at model results in a routine way without considering them from various perspectives. Starting to discuss model results in terms of probabilities over relevant time horizons can help those in the insurance and reinsurance industry maintain better perspective on the risk they assume.