Modelled expected loss figures for catastrophe bonds close to reality

Sponsors of and investors in catastrophe bonds will be pleased to read the recently published paper which looks at the modelled expected loss figures of new cat bond issuances and concludes that the numbers derived at issue are very close to the actual loss experience of the cat bond market. The paper from consulting firm and broker-dealer Lane Financial LLC shows that expected losses are close to experienced losses, meaning that in cat bonds WYSIWYG (what you see is what you get).

The expected loss is one of the metrics that every cat bond isssuance has and is derived using statistical analysis or a risk model to attempt to portray the probability of a loss of a certain size occurring on a specific cat bond transaction. The expected loss metric is often used as a way to gauge the riskiness of a cat bond by investors and as a way to assist with deriving pricing. So for us they offer a way to look at risk appetite and the level of risk in the outstanding cat bond market as well, such as in this article from February.

The folks at Lane Financial take things a step further by looking at the expected losses of all cat bonds looking back to the beginning of the market and then compare the average expected loss with the actual experienced losses that the market has seen over its history. The paper aims to uncover how accurate, or reliable, the expected loss calculations from risk models are as a way to assess the riskiness of a cat bond transaction. To achieve this Lane Financial looks at the full universe of, and history of, the ILS and cat bond market.

Lane Financial doesn’t distinguish between the different risk modellers in the paper, preferring to aggregate and average the expected loss figures from all issued cat bonds and compare them to the actual catastrophe loss experience of the ILS sector. To achieve this Lane Financial looked at 404 individual tranches of cat bond or ILS deals issued up to mid-2012, giving them a total amount of issuance to analyse of $39.6 billion. The paper itself explains the methodology of the study in much more detail.

Lane Financial take a weighted average expected loss across all 404 tranches, coming up with a figure of 1.57% using Standard Sea Surface Temperature model outputs and 1.69% using Warm Sea Surface Temperature models. So Lane Financial use a range of expected losses from 1.57% to 1.69% combined with an average coupon (or interest) payment of 7.4%, which for the $39.6 billion of issuance gives a total of annual premiums of $2.93 billion. Using the expected loss lower and upper figures of the range that would suggest that the cat bond and ILS market would have suffered losses of between $622m to $669m over the course of its history.

The paper looks at the loss history of the ILS and cat bond market, but for the purposes of the study only considers losses caused by natural catastrophe events as this is what the expected loss figures attempt to forecast the likelihood of. The total natural catastrophe losses of the 404 tranches of ILS and cat bond notes issued in the markets history comes to $682m. Lane Financial notes that while this is slightly higher than the expected range of $662m to $669m it is close enough to come to the conclusion that with cat bonds, what you see is what you get (WYSIWYG).

Lane Financial also looks at the probability of attachment, which averages 3.2% for the 404 tranches, and concludes that this is also a very good predictor of risk for cat bonds as it suggests 13 bonds should have attached and in fact just 12 did, although only 5 are natural catastrophe losses. The paper also looks at the loss ratio of the cat bond market, by comparing the premiums of $2.93 billion with the experienced losses of $682m giving a loss ratio of 23%. This is comparable to underwriters such as RenaissanceRe and Berkshire Hathaway in their better years, according to Lane Financial.

Finally, Lane Financial updated the report at the last-minute to account for the settlement on the Nelson Re bond, where arbitration was withdrawn recently. This adjusts the numbers in the study by $38m, bringing the number for experienced natural catastrophe losses to $644m, which is right in the middle of the expected loss range of $622m to $669m.

This is a great piece of research by Lane Financial and it helps to show that when you assess the catastrophe bond and ILS asset class across its entire history the numbers used to portray how risky deals are come right into line with the actual loss experience of the sector. Of course the results could be very different if you take chunks of the market and compare them, or a single years issuance, but looking at the market as a whole over its history clearly shows that Lane Financial is right when it says that WYSIWYG.

You can access a copy of this study from Lane Financial by visiting their website and registering to download the paper.