July 1, 2003 by Vikki Spencer
The 2003 hurricane season is already rearing its head – at the time of this writing, Tropical Storm Bill threatens the Atlantic coast of the U.S., with the potential to become a hurricane. In fact, the U.S. National Oceanic and Atmospheric Administration (NOAA) predicts 2003 will bring 11-15 tropical storms, 6-9 hurricanes and 2-4 major hurricanes (registering category 3 on the Saffir-Simpson scale, or higher). The reason, they say, is La Nina.
The lesser-known “little sister” of El Nino, La Nina is a 12-18 month cycle of cooling equatorial Pacific waters that occurs every three to five years. As El Nino came to an end in April to mid-May of this year, scientists said La Nina was on its way. With it comes the increased likelihood of hurricanes in the southern Atlantic or Caribbean basin, explains Amir Shabbar a climatologist with Environment Canada. Hurricanes come off the coast of Africa, west to the Caribbean and travel up the Atlantic coast. “It [La Nina] actually suppresses the jet stream over the southeastern U.S. so that any build-up of hurricane clouds is not capped.”
Certainly the southeastern U.S. is more closely associated with hurricanes than the Canadian coast. “The normal path [of a hurricane] is toward Canada, specifically eastern Canada, but before they get there, they’re traveling over much cooler water,” explains Shabbar. “By that time [they reach the East Coast of Canada], they are tropical storms – they have lost much of their strength.”
Not only is the intensity of the storm often reduced, but the vulnerability of the Atlantic provinces is less, sources acknowledge. Newfoundland, for example, has a population spread far different from the hotel-lined beaches of Florida’s Atlantic coast, explains Chris Briggs, property director, corporate underwriting at Zurich North America Canada. Maritime structures are more commonly designed to withstand severe weather, including winter weather, and the coast is a rugged, rocky one. Nonetheless, Briggs says, a summer of heavy rain could bring problems for insurers, including water damage claims, and their attendant – mold.
Briggs notes that even central Canada has not been immune to hurricane activity. The most famous is Hurricane Hazel, which struck the Toronto area in 1954, causing massive damage. Flooding was the most severe issue, and since that time the city has put in flood protection programs. Even last year, three tropical cyclones (called hurricanes in the North Atlantic/Caribbean, or typhoons in the Northwest Pacific) were registered by the Canadian Hurricane Centre (CHC). Among these, Hurricane Gustav hit Nova Scotia in September, bringing as much as 102 mm of rain and winds as high as 122 km/h in some areas, as well as minor storm surge flooding. Other storms, such as Hurricane Isidore, caused high rains in many areas of the country. Perhaps the most powerful in recent memory, Hurricane Luis, a category 3 storm, hit the Avalon Peninsula of Newfoundland with 105 km/h winds and 60-120 mm of rainfall (see chart 1). In fact, Canada is affected by 3.3 tropical cyclones per year, although this number jumps to an average of 4.9 events over the past decade, and 38-40% of all Atlantic tropical cyclones affect this country. The CHC predicts a severe tropical storm could occur in Southern Ontario sometime within the next 15 years.
The most famous of hurricane predictors is Professor William Gray of Colorado State University, who concurs with NOAA’s view that the Atlantic coast can expect increased hurricane activity this year as a result of La Nina conditions. He writes, “in six of the last eight years (1995, 1996, 1998-2001) when El Nio conditions were not present we have seen a yearly average of 9.2 hurricanes, 43 hurricane days, 4.33 major hurricanes…”.
But do insurers put any stock in such predictions? “Yes, it does hold a lot of water, especially in the U.S.,” says Briggs, but he cautions, “they can predict the level of activity, they can’t predict what that activity will be.”
The predictions are used largely for informational purposes, rather than as a basis for underwriting, adds Jeffrey McCarthy, assistant vice president and actuary at State Farm Insurance Cos. “We estimate a rate level more on a long-term basis,” he says. “What we do like is it gets it out into the public eye. It’s an awareness issue.” Attention needs to be drawn to proper building codes geared towards mitigating damage, not just for insurers’ sakes but also for governments paying out disaster relief funds.
This is no less the case in Canada, with the Insurance Bureau of Canada (IBC) noting, “disaster recovery payments by insurance companies and taxpayers have been doubling every five years throughout the 1980s and 1990s, an alarming trend that cannot be allowed to continue”. Beyond the public awareness issue, however, predictions should not be the basis for significant response on the part of insurers and reinsurers, sources agree. “It should also be stressed that the reliability of the hurricane ‘forecasts’ has proven to be moderate to poor in the past,” says Ernst Rauch, geophysicist and head of meteorological risks in the Geo Risks Research Department at Munich Re.
“All of these predictions have a degree of uncertainty to them,” comments Peter Dailey, manager of the atmospheric sciences department at AIR Worldwide. “This whole business of making predictions for the coming year is a very complicated problem which no one does very well. You might find they’re doing just as well as flipping a coin.” What the industry is increasingly turning to are “risk modeling” tools.
“Weather catastrophes – as well as earthquake catastrophes – can so far not be predicted to any reasonable degree,” says Rauch. “However, risk models allow us to assess the long-term probability of losses of a certain magnitude from different types of natural catastrophes. We consider the modeling of hurricanes or earthquakes a major step towards an improved understanding of the risk and therefore improved risk management,” he adds. Models, explains Dailey, take into account data over the course of thousands of years, and rely less on short-term climate factors such as La Nina. “We’re not concerned with what any year will turn out, but what will happen in the long term. They [insurers] are looking for the 50 and 100-year events.”
A company’s historical loss data has fallen out of favor in terms of predicting future losses, notes McCarthy. “The industry has always felt that the past is the key to the future…With the many changes in society and the potential changes in climate, this will no longer hold true,” observes Mark Baker, administrative services supervisor of State Farm Insurance Cos. In his study, “Natural Hazards and the Canadian Insurance Industry”, he writes, “there’s movement afoot, led by the reinsurance industry, and supported by research organizations like the Institute for Catastrophic Loss Reduction (ICLR), to move from actuarially-based deductive models using historic records to more predictive inductive modeling”. But, Baker concedes that these inductive models for underwriting may be a more difficult “sell” to policyholders due to their complexity. As such, the viability of models will only be proven over time, says Dailey, but insurers and reinsurers facing large losses are glad to have the tool at their disposal.
In the case of Hurricane Andrew, the worst natural disaster on record in terms of insured losses (see charts 2 and 3), risk models did prove useful, he adds. Although AIR’s original loss estimates were viewed as excessive, in the final analysis they were quite close to the actual losses incurred. On the other hand, no insurer feels prepared to wager on what “Mother Nature” might hand out, notes Briggs. “Whenever a hurricane happens, they [insurers] just keep their fingers crossed.”
There is a divergence of opinions on whether hurricanes are on the rise. There is growing evidence that frequency is on the rise over the last 10 or 20 years, says Dailey, but nothing conclusive. While
increasing frequency of some severe weather has been linked to global warming, other factors seem to be mitigating this for hurricanes.
Over the last 20 years, fewer hurricanes have struck the Atlantic coast than the century average, although the mean temperature of the sea has increased, notes a Swiss Re report on tropical cyclones. However, there is widespread agreement that the severity of the risk has increased. “If you go back into the 1920s, we had the same level of hurricanes, but there wasn’t the same population density on the coast,” notes Briggs. Some of the “killer” hurricanes of the 1920s and 1930s “would make Hurricane Andrew look quite small”.
And, observes McCarthy, “a storm making landfall 30 years ago may have had very little insured losses,” as there was not the same development along coastlines. The risk has shifted with the growth of population and the movement of people and buildings towards the coastline. “Exposures are being located more and more in more vulnerable areas,” comments Dailey.
And, with hurricanes come multiple risks: wind damage, rain water, storm surge, flood and sewer back-up. Between wind and flood, flood is viewed as the more costly risk, despite flood exclusions widely in place. “Depending on the type of hurricane, the location of landfall and other meteorological/hydrological parameters, flooding damage can make up almost 100% of the insured loss,” notes Rauch, referring specifically to Hurricane Allison in July, 2001 (see charts 2 and 3). “A severe hurricane that made landfall on the east coast would potentially cause major problems in terms of flooded and washed away homes,” notes Baker. “While direct damage from storm surge would not be covered for residential policies, business policies would in many cases provide coverage. In addition, sewer backup endorsements for both commercial and residential policies would apply.”
Models commissioned by the ICLR show that a hurricane of the level of Hazel hitting Ontario today could, even with current flood planning, cause $640 million in damage, with over $400 million of that in sewer back-up.
Most commercial and residential policies in Canada and the U.S. cover wind damage. Some large structures such as airplane hangers may face a separate deductible. Flood from river overflow is generally excluded, but sewer back-up is often covered. However, flood can be bought as a separate cover unless there is a known exposure. In the U.S., insurers and reinsurers have refused to accumulate significant exposure in vulnerable areas, but excess and surplus lines coverage is also more widely available to counteract large deductibles or low limits on coverage. With a hurricane bearing down, brokers on the U.S. coast will see their binding authority suspended until the impending threat passes.
On the question of reinsurance, insurers have to look at their portfolio and ask what level of risk they are willing to retain. Beyond traditional reinsurance, securities and bonds (cat bonds) may be an option for transferring risk.
In 1992, insurers and reinsurers were put to the test in their ability to handle the hurricane risk with the devastation caused by Hurricane Andrew. All of the risk factors were “at the extreme” in Andrew, says Dailey. The industry faced a US$20 billion price tag as the category 5 storm (the most intense on the Saffir-Simpson scale), hit large buildings and highly populated areas near the coast. “That event was a wake-up call to the industry and others as well. The magnitude, the damage it created, it became pretty clear we were underestimating what the hurricane exposure is,” says McCarthy.
Andrew spawned market disruption that led to a state-run residential catastrophe insurance program. It also highlighted exposures insurers were not prepared for, such as winds driving the salt water surge inland, destroying orange crops. Andrew is still very much on insurers’ minds today, says Dailey. “Is Hurricane Andrew a once in a lifetime event, a once in a hundred years event?” AIR has classified Andrew as a hundred-year event, not just in terms of intensity, but also where it hit and the losses it produced. “It’s not nearly as rare as people might think.”
Could the industry withstand a hurricane on the level of Andrew right now? Given capital depletion, the plague of asbestos and other long-tail claims taking a toll on insurer/reinsurer reserves, and the desperate need to turn a profit and make returns for shareholders, the answer is “no”. Briggs notes, “it’s a scary proposition. The insurance industry is very weak right now. A lot of companies are very weak compared to where they were a few years ago.”
Hurricanes are not the only threat La Nina brings. Shabbar notes Canada could see the weather trend’s effects even more clearly come the winter. “Looking at historical data, typically La Nina brings a cooler winter to southern Canada, but over the last five to ten years we’ve actually seen warming.” This may be the result of global warming, he notes. “They [La Nina conditions] also produced heavier than normal snowfalls, particularly over the prairies and over the Great Lakes area.” In 1997-1998, El Nino brought reduced snowfall that saw substantial savings for the insurance industry in B.C., Shabbar notes.
With auto insurers struggling to get that product back on track, they can ill afford a winter of heavy accidents related to snowfall. La Nina may also bring the possibility of another ice storm, adds Briggs. 1998’s ice storm cost Canadian insurers $1.4 billion. When one considers the U.S. economy is 10 times that of Canada, the ice storm ranks on the same level as some of the worst hurricane damage seen south of the border. So, when the hurricane season ends in November, insurers here may be just beginning to cross their fingers.