State regulators must exercise the same oversight of catastrophe modeling firms that they do over rate advisory organizations, consumer advocates told a National Association of Insurance Commissioners panel, citing concerns that "near-term" models that show dramatically greater hurricane risk were being used by insurers to support collusive price increases.

Characterizing the models as a "black box" that, due to assertions of trade secrets and proprietary data, most states do not currently have the ability or opportunity to review, representatives of the Consumer Federation of America and the Center for Economic Justice said the NAIC should force modeling firms to disclose their methodologies and the loss data used to compute their forecasts, and if necessary, subpoena information on the firms' contracts with insurers.

At issue during the public hearing before the NAIC's Property and Casualty Insurance Industry was the March 2006 introduction by Risk Management Solutions of a new windstorm model predicting a 40% increase in landfalling hurricanes in Florida, the Gulf Coast and Southeast relative to the frequencies that would have been predicted using hurricane frequencies from 1900 to 2005.

The problem with using a long-term means for hurricane modeling, according to Mitch Sattler, RMS' vice president of public policy, is that it failed to account for the varying periods of high- and low-hurricane activity anticipated under Atlantic Multidecadal Oscillation theory. He noted that RMS' own model underestimated Atlantic Basin activity for every year after 1994 except the three El Nino seasons of 1997, 2002 and 2006. The 100-year mean also may not properly anticipate changes brought on by climate change, he argued.

But CFA Director of Insurance J. Robert Hunter and CEJ Executive Director Birny Birnbaum argued the choices made by RMS, and subsequently by competitors Eqecat and AIR Worldwide, to shift the time horizon to five years from a longer term and to shorten the pricing return period was motivated by industry demand to justify higher rates or withdrawals from certain markets. Birnbaum suggested modelers were caught in a cycle of "reverse competition" that failed to reward the best science, while Hunter alleged insurers "broke their promises of a stable market" made when the models were widely adopted following Hurricane Andrew in 1992.

"I think it's clear that the insurers put pressure on the modelers to come up with a method of raising the price" of coverage, Hunter told the panel, citing a letter from RMS noting that in developing the near-term model, the company had "taken counsel from the insurance industry…to find the duration over which they sought to characterize risk."

"That's not science, that's politics," Hunter said. He excoriated state regulators for not doing more to question the models' use in rate filings, noting that only Florida and Louisiana have disallowed the near-term models, while Georgia has never accepted models in the rate-filing process.

Representatives of the modeling firms conceded that introduction of the newer models was driven by their clients' demand for products to assess near-term risks, but noted that those clients were not limited to insurance companies. Sattler pointed out that while an insurer might desire to show a high level of risk when making a rate application, it wouldn't have the same priorities when dealing with brokers, reinsurers, rating agencies or participants in the capital markets, all of whom also make use of the models.

"Our customers have no input into the results that are generated from the models," Sattler said. "The models are used by all parties involved in the risk transfer process."

Sattler noted modeling firms publish information about their projections, in the form of analyses on a hypothetical database, in their applications to the decade-old Florida Commission on Hurricane Loss Projection Methodology, which was established to audit catastrophe models. He said the firms would be "willing to consider" an expansion of that process to include licensing by other states, but insisted that proprietary information must be safeguarded and suggested a centralized process that would allow a certified model to be used in all states.

Insurer advocates staunchly opposed regulation of the modeling industry, with representatives of the American Insurance Association, Property Casualty Insurers Association of America and the National Association of Mutual Insurance Companies arguing that rejecting near-term models would result in artificially suppressing rates and could threaten insurers' financial solvency.

"The wisest course would be to ignore Hunter and Birnbaum's scientifically illiterate tirade against catastrophe models," said Robert Detlefsen, NAMIC's director of public policy. "It would be a mistake for regulators to assume that they can somehow determine whether a particular model is valid or useful. They cannot."

----------------------------------------------------------------------------------------

Copyright © 2007 A.M. Best Company, Inc.

Presented by InsuranceHeadlines.com