Lynda L Downs, Andrew Mitchell and M-C Papaefthimou

Modelling National and Social Hazards for the Insurance and Reinsurance Industry Worldwide


The increase in storms worldwide since the late 1980shas been a wake-up call for the insurance/reinsurance industry. The need was recognised for a more scientific approach to financial analysis with respect to environmental risk. It is imperative that the insurance and reinsurance industry continues to:


For the past four years at Greig Fester, the research and development team has been working on a long-term Geographical Analysis Project (GAP) for the estimation of catastrophic and attritional risks faced by insurance companies and their reinsurers. GAP combines environmental, demographic and insurance data to provide both portfolio management and loss scenario services. GAP received an AGI award for technological progress in 1995 and has led the insurance industry's use of GIS techniques.

This paper examines key issues of research and development at Greig Fester, an international, privately owned reinsurance broker. As reinsurance brokers, clients are insurance companies and the purpose of research is to spread better the accumulation of exposure of individual companies by arranging the purchase of reinsurance in the major markets worldwide. Therefore research and development teams within Greig Festers concentrate on such hazards as earthquakes, hurricanes as well as hazards that are more local to the UK, such as wind storm and flooding. This paper will concentrate on the issues of data storage, handling and manipulation within a model.

One way to gain a better understanding of the nature of a risk is to model it. However, it is imperative to understand that a model is only a simplified representation of a real world process. Therefore, all modelling techniques have limitations and it is important to know the limitations and associated errors. All spatial modelling is conducted in ArcInfo using the GRID and TIN modules on a UNIX platform. On some models C++ programming is incorporated, as well as S-PLUS and PC based numerical models. ArcView is used for display and demonstration.

Models

All modelling at Greig Fester is in a continuous process of development. As new and more accurate data becomes available and methods and technology become more advanced, models are updated and improved . All Research and Development teams within Greig Fester are using modelling, both geographic and numeric, to analyse the hazards most prevalent in their area. In the UK division two models (wind storm and coastal flood) are currently in the next phase of development.

Storage, handling and maintenance

Developing a model is just half the problem; data storage, handling and maintenance need forethought to avoid obstacles. A major issue in developing models that use geographic information is how to store and manipulate large volumes of data. Both the flood and wind storm models utilise vast quantities of data; for example, the digital terrain model (DTM) for the first generation flood model is approximately 4.5 gb of data. The next generation flood model will be even larger - the storage and handling can prove unwieldy. ArcInfo Librarian is being used as a tool to better manage the flood model.

Another issue is the use and maintenance of post code data. A standard way to analyse a client's portfolio is by using the Royal Mail post code system (area, district and sector post code). With the full unit post code, every residential and non-residential property in the UK can be located to +/- 100m. Presently Ordnance Survey is in the process of increasing the accuracy by locating the centroid of each full unit post code, however, maintenance of this data set becomes a problem as the Royal Mail adjusts and updates the post code system approximately four times a year. Keeping on top of maintaining post code data is a necessary chore, but equally as important is cataloging past datasets for studying past events.

Conclusion

Studies and models at Greig Fester examine existing assumptions about damage factors in the light of real insurance data, from clients, with the aim of producing comprehensive damage scales suitable for extensive loss scenarios. The approach for all modelling at Greig Fester, thereofre,is to explore the combined effects of the different hazards to which an insurance account is exposed and to produce fully calibrated damage scales, enabling the insurance industry to quantify the risks it is protecting.


Lynda L Downs
UK Division, Greig Fester International, Ltd
Devon House
58-60 St Katharines Way
London
E1 9LB

Telephone: 0171 488 2828
Fax: 0171 265 1234