The insurance industry is making strides in how well it assesses and finances earthquake risks, but there is still plenty of uncertainty surrounding these natural hazards.

Reinsurers, insurers and corporate buyers today all have better information about their exposures than a few years ago, thanks to sophisticated computer modeling programs. Yet some critics caution that these catastrophe models are only as good as the information on which they are based.

And the experience of building codes in recent earthquakes may offer lessons for improving property loss prevention in other areas of the world.

Yet when all is said and done, each new earthquake highlights additional areas for improvement and research, as the recent Italian earthquakes show.

"The progress in this area has been enormous," Werner Scharf, head of the Natural Disasters Unit at Swiss Reinsurance Co. in Zurich, said of catastrophe modeling for earthquakes and other natural disasters.

"The development of computer technology and severe losses has made insurers and reinsurers much more aware of the risk. It really is a revolution," he said. Swiss Re uses catastrophe modeling programs to model the energy released by earthquakes of different sizes to gain a picture of loss potential. "The results of our calculations are reliable," he said.

Even so, Herbert Tiedeman, a former consultant to Swiss Re with four decades of experience as a loss control engineer, sees limitations in the information that the insurance industry is working with to assess catastrophe risks.

The catastrophe information base that has been developed among some reinsurance companies is just "kindergarten stuff," Mr. Tiedeman said. "Reinsurers have to look at two points: What is the risk potential? What will be the losses?"

Scientists can answer the question of catastrophe risk potential within a certain margin of error, but reinsurers still do not have reliable databases to monitor their aggregate loss exposures, according to Mr. Tiedeman.

The calculation of earthquake risk in a specific location is an exercise in faith over statistics.

The earth's crust has a history of about 4.5 billion years, while scientifically reliable records of earthquakes began only in the last century.

Mr. Tiedeman said he does not have faith in attempts at predicting earthquake risk using a database of historical information that may bear no relation to physical phenomenon. Historical accounts of catastrophes exist worldwide but these have to be carefully assessed, he noted.

Chinese historians, for example, recorded an earthquake some 18,000 years ago, but that gives a false picture of how complete the record of Chinese seismic activity truly is, he said.

"Chinese history looks good but it is riddled with big gaps," Mr. Tiedeman said.

Furthermore, historical accounts often are tailored to the philosophies or politics of the day, he said. Medieval chronicles in Europe, for example, often depicted earthquake, floods and other natural disasters as divine retribution after the death of a monarch or a popular uprising.

The real way to manage risk in an earthquake environment, said Mr. Tiedeman, is "sound engineering."

Engineers who work on making structures earthquake-resistant are more philosophical about the state of catastrophe information.

"When designing structures such as electricity substations and transmission lines in earthquake areas, the engineer has to make the best of conflicting data on risk potential. It is very arbitrary," said Soren Berggren, a mechanical engineer at Vasteras, Sweden-based ABB Corporate Research, a division of Swiss/

Swedish engineering conglomerate Asea Brown Boveri.

Mr. Berggren said that earthquake risk assessment can vary by two or three orders of magnitude depending on the way it has been calculated.

Also, prior models may not bear up under an actual earthquake, he added. Observed ground motions at electricity substation sites after the 1994 Northridge earthquake in California, for example, were double those predicted, he said.

The design of a structure in an earthquake-prone area cannot be tested first in a laboratory, Mr. Berggren pointed out. The next earthquake in the area where it is to be located will be the test, he said.

"When the only test available is the earthquake, that's a fairly random kind of feedback," he said.

In addition to underlying uncertainty about the accuracy of risk assessment, cost remains a major factor for engineers deciding how much earthquake resistance to include in their products or structures.

There are different approaches in the electricity sector, for example, depending on whether a facility will be nuclear-powered rely on more conventional sources of generating electricity, such as coal or hydroelectric dams. Mr. Berggren said nuclear facilities are subject to far more stringent safety regulations.

"Sometimes it is not economic to make a (conventional facility's) structure 30% more expensive. It is better to take the risk of the earthquake. But this cannot be done for nuclear facilities," Mr. Berggren said.

The value of that proper engineering was proved after the 1995 earthquake that struck Kobe, Japan. Buildings constructed after 1981, when a new building code was introduced in Japan, suffered very little compared with older structures.

The Kobe earthquake registered a magnitude of 7.2 on the Richter scale and caused economic losses of $200 billion.

Insured losses were only $3 billion for the Kobe earthquake, largely because much of the risk was covered through a government-backed disaster program, rather than commercial insurance.

According to research by Munich Reinsurance Co., most of the insured losses from the Kobe quake were from property claims, with relatively small marine losses, mainly from damage to the city's port. More than 15% of the insurance payments from the disaster were in personal lines such as life, health and personal accident coverage.

According to a report on the Kobe damage by Andrew Whittaker, associate director of the Earthquake Engineering Research Center at the University of California at Berkeley, the Kobe area had been considered to be of relatively low seismic risk before the 1995 quake.

Structures built after the Japanese building code was revised in 1981-82, however, suffered very little damage.

The performance of Japan's building codes in minimizing quake losses offers significant lessons for other regions of the world, especially the United States.

"There are many similarities between the Kobe region and seismically active regions in the U.S. The City of Oakland has much in common with Kobe, namely its proximity to a major active fault, extensive deposits of soft soils and bay mud beneath many of the buildings in the downtown area, and a significant inventory of non-ductile concrete, steel and masonry buildings," the University of California's Mr. Whittaker concluded in his report.

"If there is a lesson to be learned by policymakers from this earthquake, it is that if a severe earthquake strikes a major urban area in the U.S., the social and economic losses will be huge and the impact on the economy will be significant. Improvement in model seismic design codes and the timely introduction of mandatory requirements for seismic rehabilitation of vulnerable buildings are key steps in mitigating the risk posed to the built environment by earthquake shaking," the report said.

Swiss Re's Mr. Scharf, however, said he does not believe the U.S. and Japanese building codes are too dissimilar.

"I would not say that there are large differences between Japan and the U.S.," he said. "Japan is No. 1 in terms of code requirements, and New Zealand also is very good and has a high level of engineering," he added.

The unpredictability of earthquake risks was recently illustrated by twin temblors of about 5.5 and 5.7 on the Richter scale that shook central Italy in September.

The quakes, followed by smaller aftershocks, caused 11 deaths, 126 injuries, and damaged 88,000 homes, leaving 130,000 people homeless. Valuable art treasures-most notably the Basilica of St. Francis in Assisi-were damaged or destroyed, causing tens of millions of dollars of damage, according to Italian government sources.

But insurance losses are expected to be minimal as private insurance coverage was sparse.

In the basilica, the earthquake destroyed reinforced concrete vaulting that had replaced earlier wooden vaults. The vaulting was supported by the basilica's fresco-covered walls.

After the earthquake, critics charged that concrete arches were too heavy for the church's walls to support, contributing to the damage.

But Italian restoration experts countered by claiming that when the old wooden vaults needed to be repaired, concrete structures were the best technology available at the time.

Mr. Whittaker noted that earthquake engineers are trying to strengthen a number of important historical buildings throughout the Mediterranean countries, some up to 2,000 years old. Even though many buildings on historical sites do not have foundations, they still can be strengthened by more technology, but the cost is high.

"It's always a question of money. You can provide any elegant solution if you throw enough money at it," he said