Lis Gibson and Kendra Felisky say insurers must take care in using models

Hurricanes present huge problems for insurers. While these events are not a new phenomena, in the past two years hurricanes have hit at an unnerving rate.

The big question for the global insurance industry is whether insurers are adequately prepared for 2006.

To be successful, an insurer needs to understand the level of risk it has assumed in its book of business. The consequence of this is that the insurer must price its business properly for the level of risk that it is assuming.

In 2005, most insurers fell foul of both. They relied heavily on third party catastrophe models for both pricing and monitoring their aggregate exposures. When these models proved to be inadequate, the insurers suffered.

Keeping control of aggregate exposures is key to being a successful insurer. A poorly controlled insurance company will only be concerned with income and profit, not with the level of 'riskiness' in their portfolio or understanding the dangers of concentrating its business in one geographic location.

Deficient models
The 2005 hurricanes showed that the catastrophe models were deficient in several ways. First of all, they underestimated the probability of a large hurricane like Katrina. Secondly, the models did not cover all the exposures that could arise from such a large hurricane.

But these deficiencies were not all the fault of the catastrophe modellers. The model users are to blame as well for assuming that all exposures were being covered and not thoroughly checking what the models included and excluded.

So, for 2006, have insurers adjusted for the deficiencies with the catastrophe models when pricing business? Each insurer needs to decide whether to rely exclusively on the information from the catastrophe models or whether adjustments such as a loading are needed.

The second question is whether insurers are better at controlling their aggregates in view of what happened with the 2005 hurricanes.

Many insurers added a loading similar to that which they added when pricing the business. It is fundamental for insurers to monitor their exposures as they are writing the business. Calculating the aggregates a month after all the business has been written is not helpful.

And tight control of underwriters who are over-utilising their capacities is important.

It is also important to thoroughly consider where the exposures may arise.

After the 2005 hurricanes, insurers had losses arise from areas of the book that were not included in their aggregate exposures calculations, but if they had thought about it for a bit, they might have realised that there were potential exposures.

The timing of the renewal season is also important. Most catastrophe business, at least the North American portion, renews on 1 January, which means that insurers would have been assessing the exposures in November and December.

The deficiencies in the catastrophe models mentioned above were suspected but the full extent was unknown.

If the insurer made appropriate adjustments to the models, then they probably charged an appropriate price for their risk and have modelled their aggregate exposure correctly for 2006 catastrophes. If inadequate adjustments were made - or none at all - the insurer could be disastrously exposed to a catastrophe in 2006.

Rates and policy coverage must also be considered. Rates did increase for property catastrophe business for the 1 January renewals. But did they increase enough? If insurers made some adjustments to their catastrophe models, but not enough, then they could still experience significant losses.

Another way of monitoring their exposures is to reduce the business written in over-exposed areas or in certain types of coverages.

Some insurers have reduced cover dramatically in areas where they turned out to be overexposed, or in areas where they didn't feel confident that the exposures could be modelled appropriately. A prime example of this is in offshore energy where the programmes for the rigs have been dramatically restructured after the 2005 hurricanes.

Energy losses
If 2006 is another historically extreme hurricane year, it seems unlikely that markets will be so forgiving.

It is believed that $15bn of energy insurance losses have been inflicted over the past two years in the US Gulf of Mexico from

hurricanes. This compares with annual premiums in the region of just $500 million a year. Clearly this creates a major issue for the viability of the current insurance structure.

Insurers reacted by increasing premium rates and significantly decreasing insurance coverage. But if current levels of hurricane activity continue, recent adjustments will not be enough to reset the balance.

It might be a case of three strikes and you're out. Another damaging hurricane season shouldn't be such a bad event if insurers have learned their lessons and kept tight control of their exposures.

Conversely, if 2006 is a benign year for hurricanes, the risk for the insurance industry is that the lessons and Katrina and Rita will quickly be forgotten.

No matter what the 2006 hurricane season is like, insurers must now realise that they alone are responsible for monitoring their aggregate exposures and that they cannot blindly rely on vendor catastrophe models to do everything for them.

Without a substantial change to the way catastrophe modelling is used, there is a very real risk that hurricanes could become the next asbestos for insurers.

But, unlike the drip-feed of asbestos liabilities, hurricanes happen in two days, not over two decades, and capital providers and ratings agencies will be keen to see that insurers have taken steps to improve their risk exposure. IT

' Lis Gibson and Kendra Felisky are with the insurance practice at Deloitte

Weather theory
There is a theory which is gathering support across the insurance industry that the Gulf of Mexico hurricane activity, especially in relation to the frequency and severity of hurricanes making landfall, varies with a macro weather cycle.

In the first part of the 20th century, landfall hurricanes were more frequent and more severe than in more recent times. However, in the early years of the 21st century, Gulf water surface temperatures have increased and this may signal a return to the older pattern of hurricane activity.

During the more benign period of activity, in the past 30-50 years or so, there has been extensive development of coastal areas which are now in the front line for any new severe weather phase.

Add to that the possible impact of carbon emission global warming, and the next 20 years might suffer repeated and severe hurricanes with ever growing insurance costs.