S&P: Using multiple models would increase transparency in the catastrophe bond market

Ratings agency Standard & Poor’s has issued a statement on the topic of using risk models from multiple vendors to assist in the rating of catastrophe bonds. Back in April, S&P put 15 catastrophe bond ratings on CreditWatch negative due to the introduction of a new version of the RMS U.S. hurricane risk model. The new risk model was expected to impact the probability of attachment of these bonds so S&P requested that they all be run through the new RMS model to see how it changed the view of risk for each catastrophe bond.

While the cat bond tranches were assessed again by RMs using their new model, S&P also had many of them assessed using the AIR Worldwide hurricane model as well. This allowed S&P to compare and contrast the modelling results and take both sets of results into account when considering whether they needed any change to their ratings.

The ratings for the catastrophe bond tranches which were affected by the model change have no all been resolved by S&P, with six downgraded in the first batch that were assessed and five more downgraded by the end of July.

S&P rely on modelling from the three main catastrophe risk model vendors (RMS, EQECAT and AIR Worldwide) for all rated natural catastrophe bond transactions.

S&P say that despite all cat bond transactions to date using a single risk model they advocate the use of multiple models and they say they regularly express an interest in looking at multiple models for ratings in discussions with sponsors, risk modelling firms and investors. S&P ratings criteria allow for the use of models from multiple vendors, however they will continue to rate cat bonds based on the output of a single vendor risk model.

S&P say that using multiple models for assessing the risk of cat bond transactions would increase transparency in the market. They say that the risks are typically low-frequency and high-severity. Probabilities of attachment or loss and actual loss estimates can diverge significantly from one risk model to another according to how the data is interpreted, say S&P. They therefore say that a multiple-model approach would give investors (and sponsors) a better view of the range of potential outcomes. S&P close by saying that while this would not eliminate uncertainty, it should provide greater insight into the actual risk of a cat bond transaction.

The statement from S&P is not unexpected and makes sense for them to call for the view of risk of a cat bond transaction to be better understood. Perhaps they should lead the way by insisting that deals they rate are assessed using the three most available risk models?

Recent events are also showing that the rating itself is not essential for an investor to be comfortable with the risk presented by a cat bond. The risk modelling enables the real probability of loss to be understood and to some investors is considered more important than the resulting rating, from someone like S&P, itself. The recent spate of private catastrophe bond transactions which have come to market without a rating agency assessment shows that investors are willing to take on cat bond risk without a rating in place. Investors are happy enough after seeing the modelling output, probability of attachment and expectations of loss that they don’t always feel a rating is essential to commit funds to a transaction. However, ratings are of course essential for some sophisticated investors to be able to invest in a deal due to regulatory restrictions.

The question of whether multiple models should be used is one that is likely to be a hot topic over the next few months as the market adopts and comes to terms with the new RMS U.S. hurricane model. Whether a multiple-model approach could become the norm for issuers, we’re not so sure as the cost overhead of using all three could put off some sponsors. Investors however are bound to take advantage of as many modelling tools as are available (and they can afford) to assess their catastrophe bond positions.

RMS advocated a multiple vendor approach to risk modelling in April (our article here). Willis also suggested that some investors will look to utilise multiple models when performing due diligence on a cat bond deal.