The computers at Shah's Silicon Valley company, Risk Management Solutions Inc., contain mathematical models of every U.S. disaster from the 1812 earthquake that toppled chimneys in St. Louis to the 9/11 assault that brought down the twin towers in New York, as well as 100,000 synthesized "extreme events."
RMS runs its disasters through your community — and sometimes right through your home — to see how you'd fare in a hurricane, hailstorm, earthquake, epidemic or terrorist attack. The firm sells its knowledge to insurance companies to help them decide whom to cover and how much to charge.
Since Hurricane Katrina last year, those decisions have been running pretty much in one direction.
Based in part on RMS' predictions, companies like Allstate Corp., the nation's second-largest property insurer, have gotten out of some lines of coverage altogether. It and other companies have spent the year dropping or paring back policies from Oregon to New York.
Figures from state regulators show that more than 1 million homeowners nationwide have had to scramble to land new insurers or learn to live with weakened policies. Tens of millions are likely to face rate increases between 20% and more than 100%, according to regulators. And this may only be the beginning.
"Between hurricanes along the East and Gulf coasts and earthquakes along the West Coast, it is an open question whether the private insurance industry will continue to insure the coastline at all," said University of Pennsylvania economist Howard Kunreuther, one of the country's foremost authorities on disaster.
RMS is at the vanguard of a technological revolution that's reshaping the nation's $626-billion property casualty insurance industry. The industry, once synonymous with green eyeshades and airless statistics, is embracing a new generation of powerful computer techniques to learn everything it possibly can about you — or at least people very much like you — your health, habits, houses and cars. It is using this new trove of data to replace traditional uniform coverage at uniform rates with an increasingly wide array of policies at widely varying prices.
Industry executives say the aim is to create a finely tuned system in which companies can better manage the risks they bear while consumers can more carefully pick the protection they need and pay just the right amount for it.
As insurers become more adept at the techniques, "American consumers can be more assured that their companies will be there when they need them to pay their claims," said Robert P. Hartwig, chief economist of the industry-funded Insurance Information Institute in New York.
The new techniques are already paying off for insurers. Despite Katrina and a string of other big storms last year, industry profits hit a record $40 billion-plus. With the luck of no major storms so far this year, profits are in line to reach as much as $60 billion.
But some regulators, economists and consumer advocates contend that the industry's growing use of sophisticated computer-aided methods is producing side effects that could undermine the very nature of insurance.
Traditionally, insurance companies group people facing similar dangers into pools. Company actuaries determine how often events such as illnesses or accidents have befallen pool members in the past and how costly those occurrences have been. Insurers set their rates based on the frequency and loss histories assembled by their actuaries.
A key characteristic of this approach is that there's an incentive for insurers to assemble pools as big as possible. The bigger the pools, the more the actuaries have to work with. And the more they have to work with, the more accurate their frequency and loss numbers.
But the question has always hung in the air: What if insurers could know more in advance? What if they could outflank their history- and numbers-bound actuaries and predict who's more likely to be hit with setbacks in the future? What if they could charge such customers steeply higher rates, or avoid them altogether? Wouldn't that boost profits, making shareholders and executives happy, and ensure that insurers had plenty of cash on hand to pay the smaller claims of the safer customers?
That is the promise of catastrophe models like RMS'. And it's the promise of new "data-mining" methods that let companies use a person's income, education or ZIP code to predict future claims. That in turn encourages insurers to raise rates or refuse coverage for the very people who need it most — low- and moderate-income families, for example, or those who've suffered such setbacks as unemployment.
As the industry expands its ability to "slice and dice" customers and applicants, Texas Insurance Commissioner Mike Geeslin, among others, worries that "the risk-transfer mechanism at the heart of insurance could break down."
If that happens, Geeslin warned, "insurance will stop functioning as insurance."