American insurer Allstate will no longer offer earthquake cover in an attempt to limit its exposure to rare catastrophes.

As the current hurricane season gears up for what many predict will be another violent spell, the eyes of the insurance and reinsurance community are firmly focused on the well-weathered Gulf of Mexico. With current estimates of last year's insured losses from Katrina expected to total nearly $40bn, it is hardly surprising that many of the leading insurers and reinsurers are carefully assessing their loss potential for the new season and preparing themselves for the worst. But in amongst all the speculation and debate surrounding the most prominent natural catastrophe risk, there are stirrings of concern in a rarer field – earthquakes.

The US' second largest auto and home insurer, Allstate, recently announced it would not be endorsing its consumer earthquake policies when they come up for renewal. Citing the fact that due to their sporadic nature earthquake models are inherently sparse in accurate details, Allstate believes the risks are too high for cover to be an economic feasibility. The decision will affect most of the company's 407,000 customers who have taken out optional earthquake insurance. Only those in states such as Kentucky, where state law mandates that insurers offer earthquake policies, will not lose their cover.

The action has caused some concern in the insurance and reinsurance industry, with some interpreting the move away from high-risk catastrophes as the first stage in an industry-wide simplification of risk portfolios. Allstate readily admits that the move is a mechanism to reduce its exposure to rare, catastrophic events after Katrina cost the company $1.55bn in last year's third quarter – its biggest quarterly deficit as a publicly-traded company. “We have 17 million customers nationwide. We want to be there for millions of customers when they need us most,” said Julie Capozzi, an Allstate spokeswoman. “Limiting our exposure to mega catastrophes like hurricanes and earthquakes makes sense.” Capozzi cited experts who believe there's a 62% chance of a major earthquake hitting San Francisco in the next 30 years. An earthquake of the level of the 1906 disaster would result in $129bn in insured losses.

So what does the future hold for the science and economics of earthquake insurance? The modelling agencies have sophisticated software programmes analysing the phenomena around the world. However, the science is still in its infancy and can only provide an estimated picture of earthquake activity. Considering around 5,000 earthquakes can be felt each year in the US alone, with the costliest quake in recent history occurring in 1994 at Northridge, California, which measured 6.7 on the Richter Scale and caused $20bn in damage, the technology is not yet developed enough to placate the nervous insurer.

“As the complex nature of earthquake phenomena, regional tectonics, and the dynamics of fault interaction are better understood, seismic hazard models are becoming more physically based,” maintains Dr Jayanta Guin, vice president for research and modelling at AIR Worldwide Corporation. “The seismological community is actively researching the possibility of integrating the results of dynamic models of regional tectonics and faults with those of kinematic models, historical/instrumental data, and stochastic fault-cascading scenarios to produce a comprehensive regional hazard model for California.”

Meanwile, Allstate has purchased $600m in reinsurance to insure against earthquake and earthquake-induced fire damage that might occur before the coverage ends, Capozzi said.