The population living in the hurricane exposed areas of the US is projected to increase 15.3 % by 2025 with potentially an even greater increase in exposure for insurers. John Kollar explains.

The hurricane season in the US officially began on 1 June, exposing insurers and millions of policyholders living along the coastal regions from Florida to New England and the Gulf Coast states from Florida to Texas to devastating property losses.

Weather experts forecast 11 named storms and six hurricanes (two of them intense) in the Atlantic basin this year, and there is a 63 % possibility that one or more hurricanes will make landfall in the US.

In recent years, hurricanes have caused record losses for US insurers and policyholders. Hurricane Andrew in 1992 was a watershed event for US property insurers, causing $19.6bn insured-property losses, inflation adjusted to 2001. But if Andrew were to hit in downtown Miami today - just 25 miles from where it struck ( property losses could easily exceed $53bn in inflation-adjusted dollars, according to estimates by AIR Worldwide Corporation (AIR), the catastrophe and weather risk modeling firm.

From 1949 to 2001, a total of 78 hurricanes made landfall in the continental US, 31 of which were "intense" ( that is, with sustained winds exceeding 100 miles (167 kph) an hour. Put another way, 26 % of those storms made landfall, and 40 % of those making landfall were intense.

Increasing loss severity
Over this period, catastrophic hurricanes caused insured-property losses of $37.9bn in the US or an average of $715m per year, according to Insurance Services Office, Inc.'s (ISO) Property Claim Services (PCS) unit. Adjusted for inflation, insured losses from catastrophic hurricanes during that period totaled $56.9bn or an average of $1.1bn per year.

However, Between 1989 to 2001, losses from catastrophic hurricanes reached record levels, averaging $2.5bn a year or $3.1bn, adjusted for inflation through 2001. That figure is almost eight times the $400m adjusted average annual loss for the period from 1949 to 1988.

Taking inflation into account, four of the five most costly hurricanes of the past half-century , Hurricanes Andrew, Hugo, Georges and Opal, occurred in the 11 years from 1989 to 1999. During 2000 and 2001, hurricane activity in the US was extremely mild when no hurricanes made landfall. However, Tropical Storm Allison caused $2.5bn in damages in 2001. (For further discussion of tropical storms, see page 16.)

Inflation, demographic trends and increases in per capita wealth have caused hurricane losses to rise. More and more people are moving into coastal areas that are subject to hurricanes. They are building more houses and other structures. And the properties are worth more than in the past, both because of inflation and because many properties are larger than ever before.

Based on demographic data and population growth forecasts in the latest US Census Bureau report, the US population in the 19 hurricane-exposed states and Washington, DC is expected to increase 15.3 % from 137.3 million in 2000 to 158.2m in 2025 (Table 1). In the 22 years from 1980 to 2001, four US East and Gulf coast states ( Florida, North Carolina, South Carolina and Texas) sustained nearly 85 % of insurers' $40bn tab (Table 1 page 15), from hurricanes and tropical storms, a total of almost $34 billion in insured-property losses in inflation-adjusted dollars.

Between 2000 and 2025, the population of Texas is expected to rise 27.5 % to 27.2m in 2025, closely followed by Florida at 26.3 % growth to 20.7m in 2025. In North Carolina, the rise is estimated at 14.2 % to 9.3m in 2025 and in South Carolina, it is 14.3% to 14.6m. (Table 2 page 15).

Catastrophe exposure
Previously, insurers generally relied on historical data to assess potential losses. That information had serious limitations. Insurance records do not contain enough experience to provide actuarially credible information about potential hurricane losses. However, if hundreds of years of insurance records were available, the information would not reflect current conditions. Changes in land use, population densities, construction techniques and materials, building codes and a variety of other conditions would make the old loss experience almost useless for predicting future losses.

Computer-based catastrophe modeling can provide insurers a substitute for hundreds of years of loss experience, all based on the current set of insured properties. Catastrophe models include scores of databases and computer programs to analyse information about the frequency and severity of catastrophes and to produce estimates of potential losses under various scenarios.

A typical hurricane model uses publicly available information and proprietary research to estimate the probabilities of hurricanes of given intensities at given locations and the amount of damage those storms might do. Based on meteorological data for the past century or more, a model can simulate all the possible hurricanes that could hit a location. Such a model can also estimate the likelihood of each hurricane striking in a one-year period.

Hurricane models combine information about potential storms with engineering studies about the damage that storms of various intensities cause to different types of structures. An insurer might use information from a hurricane model together with data about its own book of business to estimate potential losses and the associated probability distributions.

Financing catastrophe risks
Demographic projections and modeling indicate that insurers and society as a whole may one day face a catastrophe - a hurricane or earthquake, for example - that eclipses by several orders of magnitude the losses of Hurricane Andrew.

Aware of the possibility of unprecedented catastrophe losses, insurers and entrepreneurs have been devising ways to take advantage of the vast potential of the capital markets to spread catastrophe risk by packaging the risk as securities they can buy and sell.

Since 1995, investors have committed about $6.4bn to various vehicles for securitising catastrophe risk. Though that figure is small compared with potential losses - and not all of the money was available at any one time - access to the capital markets could one day play a significant role in many insurers' strategies for financing catastrophe risk.

An insurer can use computer models and information about the policies the insurer has written to determine its potential catastrophe losses and how much capital the insurer would need to finance that risk on its own.

The insurer could then compare the cost of using its own capital with the cost of reinsurance and the cost of catastrophe bonds or other instruments for securitising risk.

Capital-market investors can also use catastrophe models and exposure data to analyse the benefits of securitising catastrophe risk. In addition to the profits an investor might earn, securitising catastrophe risk offers another potential advantage - an opportunity to reduce portfolio risk through diversification.

The forces of Nature are unpredictable. Insurers have no control over where people choose to live, when catastrophic hurricanes or thunderstorms strike and how devastating the losses are. What they can do is manage the geographic concentration of their exposures through more efficient underwriting, use advanced tools like computer models to assess potential property losses under various scenarios and tap alternative risk transfer mechanisms for securitising catastrophe risk. With prudent risk management, our industry can cope with Nature's fury, even though we will probably never fully comprehend its mysteries.

By John Kollar

John Kollar is Vice President - Consulting and Research for Insurance Services Office, Inc. (ISO), Jersey City, NJ. He is a Fellow of the Casualty Actuarial Society. ISO is a leading source of information, products and services related to property and liability risk. www.iso.com