The Intelligent Quarterly from the publishers of The Insurance Insider

Spring 2013
 

A nasty surprise

Insurance-linked security (ILS) fund managers were left "licking their wounds" after RMS released cat bond remodelling statistics under its new Version 11.0 hurricane model. RMS told the market to expect an uptick in its view of hurricane risk under the new model but a number of investors said that the extent of the increase caught them by surprise.

However, the ILS market adjusted quickly to the news and secondary trading prices on cat bonds remained stable in the lead up to the hurricane season.

Expected losses will rise 90 percent on average under the new model for outstanding hurricane-exposed cat bonds. This means a bond with a 1 percent expected loss under previous RMS models would now have a 1.9 percent expected loss on average.

This figure was calculated using medium-term hurricane frequency rates - using long-term averages the increase was a more modest 30 percent.

For industry loss bonds, expected losses will rise as much as 2.5 times under the remodel on a medium-term basis. This is because changes to industry exposure data had the biggest impact - driving up expected losses under the new model.

In contrast, indemnity type bonds will see less of a change, with a previous 1 percent expected loss bond up to an expected loss of 1.5 percent on average.
Similarly, expected losses on coastal-focused bonds actually went down under the new model.

RMS Risk Markets managing director Peter Nakada acknowledges the new data will be "hard to digest", but says the firm stands by its revisions. "We believe that this is a better view of risk," he says.

The firm also released a set of a benchmark bonds that it has created with the help of ILS investor Fermat Capital Management. The bonds are designed to provide insight into the differences between catastrophe models.

Click to enlarge RMS says it hopes fellow modelling agencies will pick up and use the benchmark bonds.

Regardless of what other modelling firms do, ILS fund managers must now decide how to incorporate the RMS model changes to reflect current risk in their portfolio.

Capital market reinsurers and ILS asset managers are generally held to be more reliant on catastrophe modelling than traditional underwriters, though some firms do take their own independent view on risk.

In theory, higher expected losses should mean lower bond prices on the secondary market and higher prices on new issuance.

"Something has to happen," says John DeCaro, founding principal of Elementum Advisors. "Investors who have ratings criteria in their portfolio or other portfolio constraints about expected losses may be forced to sell bonds, creating opportunities for those who have a different view of the risk."

But Aon Benfield Securities ILS trader Jason Bolding says prices on the 16 RMS-modelled wind bonds only dropped by a small amount after the RMS analysis came out.

Strong investor demand for ILS was supporting prices, Bolding says.

Meanwhile, ratings agency Standard & Poor's (S&P) was reviewing RMS-modelled hurricane bonds under the new model with potential downgrades of up to three notches as a result of the changes.

But RMS vice president of risk markets John Stroughair says ratings agencies should take all credible modelling views into account. "We think all hurricane-related risk has increased - it's not the case that the 17 bonds [placed on creditwatch] are in any sense riskier," he says.

S&P director Cameron Heath says the firm considers any information a sponsor places before it when rating a new cat bond issue. "As an impartial ratings agency we can't tell a sponsor what model to use," he says.


This article was published as part of issue Summer 2011

Insider Publishing Limited - 3rd Floor, 41 Eastcheap, London, EC3M 1DT, United Kingdom. The content of this website is copyright of Insider Publishing Limited 2013. All rights reserved Insider Publishing Limited actively monitors usage of our website and products and reserves the right to terminate accounts if abuse occurs.

Π