According to Swiss Re the 1906 San Francisco Earthquake was a "milestone in insurance history" Helen Yates asks some leading industry figures what we have learned on the 100th anniversary of this catastrophe.

On 18 April 1906 in the early hours of the morning when most city residents were still asleep in their beds a massive earthquake hit San Francisco. The initial foreshock was followed 20 seconds later by the quake. It all lasted just under a minute. When the ground stopped shaking many buildings were left in ruins, at least 700 people were dead (although some suggest the actual figure is more than three times the number) and electric power lines and gas mains were destroyed. For over 270 miles the ground had literally been ripped open where the east side of the San Andreas Fault had pushed southwards. And it didn't end there. As a result of burst water mains fires were left to rage out of control for days, leading to further destruction and loss of life. At a time of unrelenting urbanisation when cities were growing rapidly, the earthquake served as a wake-up call and illustrated the potential scale of losses from natural catastrophes in highly populated urban areas.

So, on the 100th anniversary of the San Francisco Earthquake how far has the industry come in its ability to understand and cover natural catastrophes? With Hurricane Katrina still fresh in many minds now is an apt opportunity to take stock. And who better to do that than the industry itself. John Beckman, president of Carvill's ReAdvisory group; John Berger, president and CEO of Harbor Point; David Cash, chief underwriting officer of Endurance; Robert Cooney, chairman, president and CEO of Max Re; Dr Torsten Jeworrek, member of the board of management of Munich Re; and Hemant Shah, president and CEO of RMS give their opinions.

What are the most important lessons we have learned in the last 100 years in terms of covering catastrophes?

Beckman: At first blush we want to say that the industry has made great strides in the last 100 years. Policy forms have been written to distinguish between ground shake and fire following an earthquake as well as rising and falling water. Technically speaking, seismic areas have been identified, fault lines and tectonic plates have been mapped, global weather models have been created, hurricane cyclogenesis has been studied, and catastrophe software has been built. However, we still seem to be caught off guard by the more severe events, such as Hurricane Katrina.

Cooney: I think the keyword is "unpredictability" because catastrophes happen but you don't know when, where or how big they're going to be. I haven't seen any statistics, but if that same earthquake hit San Francisco today imagine what the loss would be. Maybe I'm conservative but when you talk about events that happen once in 250 years, it does seem like they happen a lot more often than that.

Jeworrek: There is no doubt that California is better prepared to cope with a mega-catastrophe today than it was in 1906 - in all respects including insurance. The insurance industry has now come to realise just how important risk management is as a means of ensuring, for the long term, that promises of cover can really be met.

Looking at the last two hurricane seasons, what have we yet to learn about insuring natural catastrophes?

Beckman: The most important lesson is to be prepared. Hurricane Katrina and the WTC tragedy have taught us that multiple lines of business can be impacted by the most severe catastrophes. We have also learned that while hurricanes make us think of wind damage, the secondary perils (ie storm surge and flood) can be even more devastating.

Berger: What we haven't learned is that having concentrations of population in catastrophe prone areas is not a good idea. The weather is great in California so people want to live there - the winters are great in Florida so people flock to the coast. If coastal people paid what they should there really wouldn't be as much coastal development.

Jeworrek: There is no doubt that the exposure has increased significantly as far as hurricanes are concerned. We must therefore continue the process of factoring this changed exposure into risk management and risk models.

Shah: It's important to keep in perspective the last couple of years. If you look at the industry's response to $100bn of market loss over two years you actually see an industry that is quite resilient and in many ways is prepared. Contrast that to 1992 and Hurricane Andrew - it was a very different picture. Since Hurricane Katrina there has been a call for recalibration of the catastrophe models. Are the models to blame and will recalibration improve the situation?

Beckman: The catastrophe models are a tool and an insurer needs to understand exactly how that tool works. It is important to note that, whatever recalibration or adjustment the cat models make, this recalibration will be to a great extent a subjective adjustment - it will be very important to understand the reasoning used by each cat model vendor to justify their adjustment.

Berger: The models are models - a hypothesis given particular weather circumstances. So while the models were criticised, I think the biggest criticism should be of the people who trusted them implicitly. What's the difference between your 1 in 250 number and your absolute aggregate? That's the number that everybody should be paying attention to because that's your room for error.

Cash: I do not feel the cat models are to blame. In my mind the cat models did and do a very good job of allowing us to identify relative attractiveness of different contracts in our portfolio. I would say a significant portion of the issues people feel exist with the cat models can be ascribed to user error. It took Hurricane Katrina to begin to crystallise that understanding.

Cooney: I think it was justified that they were criticised and hopefully the models will become more conservative, because when things go bad they tend to go bad in a big way and things that aren't supposed to correlate often do when catastrophes happen. But the models are only guiding tools - they're never going to be 100% accurate because they're based on the input of data and data is never going to be perfect.

Jeworrek: Many models are based on retrospective simulation. However, recent climate research results indicate that exposure can no longer be represented in terms of a long-term average but must be calculated on a prospective basis. Even if the necessary improvements are made, modelling capabilities will still be subject to uncertainties, and companies will have to make allowances for these and incorporate conservative assumptions into an integrated enterprise risk management.

Shah: In the wake of each major catastrophe there is a great deal to learn and that learning needs to be reflected in the catastrophe models. Our view is that there is a lot of learning necessary all around - not only on how catastrophe models are built and calibrated but how they are utilised and linked to decision making. We were the first to say that the models needed to be recalibrated in the immediate aftermath of the hurricanes of 2004 and 2005.

What type of natural catastrophe in what circumstances is, in your opinion, the greatest threat to the industry and why?

Beckman: Given the events of 2005, hurricanes are clearly at the top of many companies' agendas. However, we should not forget the potential for a major earthquake catastrophe. California has not seen a major quake since Northridge and the pressure on the fault lines in the region continues to build. But potentially the greatest threat to the industry is from multiple major events.

Berger: We came very close last year when Rita was barrelling down on Houston, on the back of Katrina. We all have vivid imaginations of a big Tokyo earthquake or a big California earthquake, a frequency of big storms, flooding in the UK. Probably more than any one individual event is the frequency of these things.

Cash: It is always the unexpected event that is the most damaging. For me it's a hurricane that skirts the eastern seaboard and makes landfall up near New York and tears up Long Island. If you're looking at peak accumulations it's got to be Miami, San Francisco, or Tokyo. Tokyo is long overdue for an earthquake event.

Cooney: I think the big one - if you were just thinking about a single event - would be a tsunami like we saw in Asia. I gather there is some geological fear about a fault zone near Portugal. If there were an earthquake there that caused a huge tsunami it would go right across the Atlantic and create huge damage to the east coast of North America. But if you look over the next 20 or 30 years I still think the biggest aggregate loss will come from windstorms or hurricanes.

What issues do you think we will face in terms of covering natural catastrophes in 100 years time?

Beckman: Over time our exposures evolve and change. Population density along the coastline continues to increase. Population density in urban areas continues to increase. Technology continues to create new exposures. The companies that understand, anticipate and plan for these changes are the ones that will survive and prosper.

Berger: The insurance industry will cope with natural catastrophes - we have in the past, we'll continue to do it, we'll adjust and new capital will come in. But I'm not sure how we cope with the political climate out there.

Cash: I believe we are experiencing global warming - I believe water levels will rise - I could imagine some devastating flood events across coastal Europe and the eastern coast of the United Kingdom. Looking a little further down the road, when you think about big catastrophes you think about large concentrations of population in catastrophe prone areas. That could very well be China in the second decade of this century.

Jeworrek: Munich Re is examining a number of scenarios and their impact on the insurance industry. These include hurricanes in the US, potential earthquakes in San Francisco and Tokyo, and winter storms in Europe. Where we will be 100 years from now nobody knows, of course. But climate change and the shift in climate zones are already noticeable today. The rise in sea levels will be reflected in an elevated exposure of coastal areas.

Shah: Well I think the most provocative issue facing the industry from a catastrophic perspective is the issue of climate. If you look at some of the science that suggests the critical changes we'll have in the coming decades I think this is going to be one of the more provocative and significant challenges for the insurance industry.

If hurricanes continue to increase in number and severity, could this alter the availability and pricing of cat-exposed lines for the long term?

Berger: Basic economics tell you that it's all about supply and demand - and clearly demand is up and supply is constant - so you're getting good rate movement there. In the rest of the world nothing is out of balance. Just because everybody would like higher prices doesn't move a market. It's when you get this disruption of that very simple supply/demand equation that things move, and we're not seeing it. The jury is out on whether certain cat-exposed lines could become uninsurable. Clearly we have had two years of increased frequency of hurricanes. If we get a third, forth and fifth year does the capital then dry up?

Cash: If we look at our business today, it is built around the concept of diversification. The idea is that if one brings different unpredictable high-severity risks together the result is a much more predictable aggregate result that can responsibly be held in a reinsurance company's balance sheet. That presumption is now being challenged by the apparent growth in the size of the US wind peril. It is almost at the point where it dwarfs the other catastrophe perils held on reinsurers' balance sheets and so begins to break down the model of diversification.

Cooney: You will always have more pressure on rates in the areas that have the losses - it's a natural reaction. In the extreme I think certain lines could become uninsurable. I don't think we're there yet but I guess if we had a couple of more years with Katrina-like events we might step back and say "hey it is uninsurable" in certain areas where the high frequency of hurricanes are. I think an interim step in going from insurance to non-insurance would be some of the alternative risk transfer mechanisms.

Jeworrek: What the supervisory authorities and rating agencies are saying suggests that there will be higher capital requirements with regard to catastrophe covers. From the insurers' and reinsurers' point of view, business potential will grow but it will be associated with high volatility and high capital needs. If appropriate precautions had been taken, Katrina would not have been the event it actually turned out to be. Proper prevention measures and risk-adequate prices, terms and conditions are good tools to tackle the question of insurability in the future.

Shah: If a Katrina-size event were to hit every year it could still remain very much an insurable peril but I think the key issue would be pricing - can adequate rates be charged to capture the higher levels of severity. If the answer is "yes" then I think the industry is innovative and resilient enough but if the answer is "no" then I think you'll see people starting to look for other solutions.

To what extent should we be putting pressure on governments to provide a reinsurance backstop?

Berger: The government isn't this magic pot of money that can just write cheques - we're already in huge deficits and somebody has to pay for that. And that's the taxpayer. Is the most efficient method to increase everybody's taxes to pay for the cat insurance for all the homeowners in Florida or is it to let the free market charge the right rates. Should the rich guy who wants to live on the beach in Florida pay for it or should the government subsidise it? The most efficient way is to let the free market insurance mechanism take care of it - not push it onto the government because the government is a prudent inefficient bureaucracy in handling these things.

Cash: As attractive as it may sound to have the government act as a stabilising agent in the catastrophe market, I feel pretty strongly that it would be a mistake, for both the marketplace and for that matter the ultimate purchasers of the insurance product. Time and time again we have seen the government involve itself in the insurance marketplace and each time, without exception, the law of unintended consequences raises its head. Each time the government sets out to save the market it ends up more dysfunctional than it was to start with.

Cooney: I think there are certain areas where you could see the government playing a role. The Florida Hurricane Catastrophe Fund for example is pretty effective for a lot of small businesses and homeowners for very low-level claims. I think the big insurers and the reinsurers are not really interested in that kind of a frequency. People have elected to live in Florida so I guess you've got to pay a cost if you want a house on the beach in Miami.

Shah: For the most part the insurance industry is providing an effective market and effective capacity for natural catastrophes and so in some cases it's a solution in search of a problem. I don't think there is a need for a public programme. We've been through this before. In the aftermath of Andrew and Northridge there were calls for some form of government role and the industry came to the same conclusion that it seems to be coming to today, which is as a marketplace we've got a pretty good solution for this. It's not perfect, we're learning and we could certainly do better but we are able to provide the market with the needed capacity at prices that enable it to work effectively.

- Helen Yates is deputy editor of Global Reinsurance

- Guy Carp/AIR develop Swiss earthquake model

- Cats could impact reinsurers' financial flexibility

- RMS ups Gulf Coast modelled losses by 40%

1906 - San Francisco Earthquake

1923 - Kanto Earthquake, Japan

1935 - Great Labor Day Hurricane, US

1952 - Lynmouth Flood, UK

1969 - Hurricane Camille, US

1976 - Tangshan Earthquake, China

1980 - Eruption of Mount St Helens, US

1988 - Spitak Earthquake, Armenia

1991 - Eruption of Mount Pinatubo, Philippines

1992 - Hurricane Andrew, US

2004 - Indian Ocean Tsunami

2004 - Hurricanes Charley, Frances, Ivan and Jeanne, US

2005 - Hurricanes Katrina, Rita and Wilma, US