Scientists make estimates for a Willis Research Network seminar

A hurricane in New York could cost from $36bn to $140bn, according to scientists, insurers and regulators.

More than 250 of the world’s leading experts were hosted by the Willis Research Network at a seminar at Princeton University, New Jersey, to discuss the challenges of modelling the risk of hurricane and storm surge in the New York Metropolitan area.

The loss estimates they created account for estimated domestic property losses alone from a direct-hit by a Category 3 hurricane, equivalent to Hurricane Katrina on landfall in New Orleans, or Hurricane Ike on landfall in Cuba.

In a worst case scenario, New York’s sea walls could be overwashed by up to three metres.

The challenges of a Northeast hurricane 'mega-catastrophe' might be beyond the scope of insurance and reinsurance alone, according to some speakers. Dr Rawle King from the Congressional Research Office in Washington DC and Franklin Nutter from the Reinsurance Association of America provided contrasting scenarios on the mix of public and private solutions to catastrophe risk financing.

Bills currently earmarked for the US Congress may emphasise an increased role for state and federally arranged catastrophe risk protection for the largest events.

Other key points from the day included:

•?Leading scientists from the climate science community, including Professor

Kerry Emmanuel (Massachusetts Institute of Technology), Dr Greg Holland (National Center for Atmospheric Research) and Dr Tom Knutson (Geophysical Fluid Dynamics Laboratory, Princeton), explored the potential of Global Climate Models and other techniques to accurately predict the future tracks, frequency and severity of hurricanes in the Northeast US.

•?Commercial catastrophe modelling companies AIR, EQECAT and RMS compared how they represent risk in industry models and integrate new scientific understanding. Considerable contrasts exist between the firms' evaluation of this risk for different locations and exposures. The conference heard that an ensemble approach may be the most effective.