Lee Coppack considers the influence of science and e-commerce on wholesale insurance and reinsurance.

Probably few people are aware of the inverse polder problem or, for that matter, the block bootstrap. 1 Yet, they may turn up in a discussion of reinsurance, particularly catastrophe risks. These are actually topics for discussion at a conference in May on the practical applications of extreme value theory and its use for effective risk measurement and management.

Insurance and reinsurance are increasingly trying to grapple with the underlying mathematics of their business. It seems curious that until Hurricane Andrew in 1992, this sort of structured approach was generally restricted to the life insurance sector. One of the consequences of Hurricane Andrew was the growth in rigorous control of aggregates supported by mathematical modelling of potential catastrophe losses.

Since then, however, the models have not been put to serious stress testing; there had been comparatively few serious catastrophes until the twentieth century departed not with a whimper but a bang. Two powerful cyclones, Lothar and Martin, swept through France and the rest of Europe, less than 72 hours apart, leaving 140 people dead and around euros 11.5 billion in total damages. Insured damages are estimated in the range of euros 5-6 million. The two European storms give catastrophes modellers and their clients in the industry the chance to examine their assumptions and the underwriting decisions made on the basis of their models in the context of a real event.

In isolation, the effect of Lothar and Martin is not expected to be dramatic. There were already indications that the downward momentum in catastrophe rates had slowed, if not actually reversed, as the bi-annual Paragon catastrophe index shows (see adjacent table) in a process which the contraction in the retrocession market will strengthen. The impact of the storm claims may help continue the process of consolidation already well underway, as some smaller, less well capitalised insurers are absorbed by bigger companies. It is more likely that the losses will be cumulative in combination with other factors, particularly any significant catastrophes.

Changing directions
Wholesale insurance and reinsurance business are at the confluence of different streams of influence, as the Global Reinsurance roundtable on e-commerce (report starting on page xx) reveals. A number of issues emerged from that discussion which brought people from the (re)insurance market together with IT people who specialise in wholesale and reinsurance solutions. Some of the thoughts drawn from their discussion are these:
Pricing is likely to become more transparent if only because that is what customers have come to expect from using the internet to comparison shop other things, including retail insurance.

For a more rigorous approach to risk exposure to be translated into good underwriting depends on the quality of information which the reinsurer receives from the cedant.

As a result, there will be pressure for increased standardisation. A corollary to increased standardisation is likely to be greater unbundling of risk. Those elements are most ameniable to standardisation and transparent pricing which will be the most attractive for repackaging as asset backed securities.Counter offers could create a more efficient market in which processing costs were a material factor in competitiveness. Typically banking has simpler processes than (re)insurance. In any case, if greater sophistication and transparency narrow the spread of pricing in conventional underwriting, costs become an even more important element of competition. Would those who were not able to participate in this market be left to compete for a diminishing book of the business which is unprofitable or difficult and, therefore, expensive to service?

Over time, e-commerce is expected to even out price competition. The next stage will be competition by service, which will in turn reach greater consistency, leaving the brand the ultimate area of saliency.

For a start up operation, the absence of legacy systems of varied compatibility would be an advantage, but our lunch guests felt that the need for business relationships and complexity of the information handled by reinsurers make Reinsurance.com unlikely in the next two or three years.However, the lesson of the Bermuda market is such that it would not be healthy for established companies to be over-confident about this. If Reinsurance.com responds to the perceived needs of the clients for price, service or convenience - and there is evidence that clients are pushing in this direction - then there is plenty of capital available. New relationships can be created and surprisingly quickly, as the Bermuda market shows. However, they may not have to be new; a web search on the url: www.reinsurance.com brought up the Cologne & General Reinsurance site.

Lee Coppack is editor of Global Reinsurance. E-mail: lee.coppack@globalreinsurance.com.

1 Polders are low lying land reclaimed from the sea and protected by dykes. James Orr, one of the authors of our article on extreme value theory on page xx has kindly supplied the following explanation: “The required dyke height is usually expressed as a likelihood of failure (i.e. waves coming over the top). That likelihood might be expressed as a probability (say, 0.01%) or as a ‘return period' (During one year in 10000 identical years). However, one might only have 40 years of actual wave height data to make that judgement! Hence, we need to extrapolate outside our actual experience. That is where extreme value theory comes in.Bootstrapping involves using observed data as the distribution from which future observations might emerge. So if you had rolled 1,3,4,5,3,6 with six throws of a dice you would (in bootstrapping) randomly sample from 1,3,3,4,5,6 (equal probability for each value) to simulate future observations. Clearly, in this case, we know certain facts about the behaviour of a dice and would normally simulate from 1,2,3,4,5,6!”

Note: The conference on practical applications of extreme value theory is organised by Risk Training in London (8-9 May) and New York (18-19 May).