Genesis of the Risk Prediction Initiative

The large losses resulting from Hurricane Andrew put many insurance companies out of business or in difficult straits, and a significant fraction of capital available for use by the insurance industry was used to pay claims. The resultant depleted pool of capital available for insurance drove up reinsurance prices and increased the potential returns for companies offering reinsurance.1 A number of new property catastrophe reinsurers were formed in Bermuda to take advantage of this business opportunity.2

In general, these new companies, as well as other reinsurance and insurance companies that survived the large losses, were highly motivated to learn more about their exposure to risk from natural hazards. The RPI was created in 1994 as a result of discussions between BIOS scientists and individuals in these companies who were seeking novel approaches for understanding natural hazard risk. The goal is to support research on natural hazards and to transform the science into knowledge that sponsors can use to assess their risk

1 Reinsurance is essentially insurance for insurance companies.

2 This cycle of events was repeated after the large losses associated with the terrorist attacks in 2001 and after the losses during the 2005 hurricane season.

(Malmquist, 1997). The October 2005 workshop was motivated in part by the active 2004 hurricane season and was designed to meet RPI's goals, but it was received with extreme interest because it was closely preceded by Hurricane Katrina and two publications (Emanuel, 2005; Webster et al, 2005) that raised the specter of greatly enhanced losses due to an increase in the intensity and number of the strongest tropical cyclones.

A number of leading companies active in the property catastrophe insurance industry have sponsored the RPI through restricted donations to BIOS, a US not-for-profit 501c(3) corporation. The companies that sponsor RPI have changed through the years in response to changes in market conditions, mergers, changes in business strategies within individual companies, and the formation of new companies.

Sponsors ofRPI represent different sectors ofthe insurance community that are active in property catastrophe reinsurance. The 2006 sponsors represent companies that are active in the primary insurance market (State Farm Fire and Casualty Company), in catastrophe risk modeling (Risk Management Solutions), and in catastrophe reinsurance (XL Capital Re Ltd., Renaissance Reinsurance Ltd., AXIS Capital Holdings Specialty Ltd., PartnerRe, Aspen Insurance, Flagstone Re, and Arch Reinsurance Ltd.).

Companies sponsor RPI for a range of reasons. Some companies hope that RPI research results will improve the catastrophe risk models that are used to estimate probable loss distributions. Other companies build their own proprietary versions of the commercial risk models and use RPI research results for their own model-development purposes. Some companies support RPI as a way of meeting their company's charitable goals. Regardless of a company's interests, their satisfaction with RPI's activities is assessed each year when it is time for the annual renewal of a company's sponsorship of RPI.

An early issue tackled by RPI sponsors involved the identification ofresearch topics that would be supported through RPI funding. The decision to focus on tropical cyclones was made for three major reasons. The first concerned the business relevance ofmeteorological hazards in general and tropical cyclones in particular. Meteorological hazards are responsible for a large portion ofinsured property losses, and hurricanes are the leading cause of insured loss (Table 16.1). The recent large losses from landfalling US hurricanes in 2004 and 2005 confirm the rationale supporting the decision to focus on hurricanes.

The second reason for focusing on tropical cyclones was the fact that the ocean is a major mover and reservoir ofheat, and heat from the ocean powers tropical cyclones. The BIOS has the longest record of ocean measurements in the world through the Panulirus Hydrographic Stations, and BIOS has strong connections to the ocean climate community. The production ofoceanographic

Table 16.1. Distribution of top 30 insured losses by hazard for the period 1970-2005

Loss (billions 2005

Hazard type

US dollars)

Percent of total

US hurricane






European wind



US earthquake



Japanese typhoon



US tornado and hail



European flood



Japanese quake






Data from Zanetti and Schwarz (2006).

Data from Zanetti and Schwarz (2006).

data and access to climate scientists provides BIOS with close connections to scientists with tropical-cyclone-related knowledge.

The third reason is related to the difference in our ability to forecast hurricanes and earthquakes. Earthquakes cannot be predicted in ''real time,'' and as a result much of the federal research support for seismology is directed towards basic research. In contrast, hurricanes can be predicted in real time and a large fraction of federal support for hurricane research is aimed at improving real-time prediction. However, most insurance-related business decisions related to hurricanes are made on annual or longer timescales, so this federal support is directed towards problems on timescales that do not match those of the insurance business cycle. In addition, funding that RPI could provide to the seismological community would have trouble rising above the noise of federal support for earthquake research. In contrast, RPI funding could produce a significant signal in the hurricane research community interested in problems relevant to the insurance community. Thus, the combination of hurricane losses and potential influence within the research community resulted in RPI's decision to focus on aspects of basic hurricane research that have business relevance.

Continue reading here: RPI activities

Was this article helpful?

0 0