As is to be expected after a natural catastrophe event with the size and reach of hurricane and post-tropical storm Sandy, the estimates of insurance industry losses keep rising. This is typical of a complex, never seen in recent times, catastrophe event. For a great example of how loss estimates can creep under these circumstances one only has to go back to hurricane Katrina, where preliminary estimates said $10-$25B, but that soon rose to $20-$35B and quickly increased again to $40-$60B.
And that was just the estimates of insured losses. The actual total from hurricane Katrina is thought to have been somewhere around the $71 billion mark, a massive 305% increase from the mid-point of the preliminary estimate up to the final total once all claims had been made.
So, look at Sandy. The preliminary estimate had insurance industry losses pegged at somewhere around the $5-10B mark. That was upped to between $7-$15B and then again to between $10-$20B. Taking the mid-point of the preliminary loss estimate and scaling that upwards by 305% (as happened in the case of Katrina) takes you up to $30.4 billion. Is it feasible that Sandy’s insured losses could reach that level and what would that mean for reinsurers and also catastrophe bonds?
So, the latest insurance industry loss estimate to come out is the first estimate from risk modeller RMS. This hasn’t been publicly announced yet, but it has been widely reported as the figure that RMS have been telling their clients. RMS has told clients that they expect Sandy to cause between $20 billion and $25 billion of insured losses. This is the highest risk modeller estimate of insured losses from Sandy yet.
So far we have AIR Worldwide quoting $7 billion to $15 billion as the insurance industry loss range, EQECAT quoting $10 billion to $20 billion and now RMS quoting $20 billion to $25 billion. Reinsurers, and also cat bond investors, must be watching and wondering where this will end.
It’s still relatively early days in a complex catastrophe loss scenario for the total extent of damages to be fully understood. Much of the estimation process that the risk modelling firms undertake relies on modelled outputs of windstorm scenarios software using exposure information from insurers to estimate the extent of losses. Physical assessment of losses, actual on the ground work, is also factored in by this stage of their analysis helping them to better understand the split between factors such as wind vs water and commercial vs personal lines losses. So the estimates increase in accuracy as time goes by, but that is normally an upwards trend and rarely downwards, particularly when the impact of a catastrophe is something never seen before and when business interruption has to be taken into account.
There are so many unknowns in the aftermath of hurricane Sandy that we find it very hard to believe that the final total won’t be higher than any estimate at this still early stage. Whether we will see insured losses scale up by 300% as happened in the case of Katrina, is impossible to predict, but $30 billion does not seem an unreasonable figure at this time. Reports suggest that Sandy will cause over 1 million insurance claims, and given the high value region that was exposed it doesn’t take a mathematician to realise that a million claims could result in a very substantial final insurance industry loss total.
So as the insured loss estimate rises more reinsurance programmes, instruments such as industry loss warranties (ILWs), county-weighted ILWs (CWILs) and of course catastrophe bonds become potentially at risk. The $20 billion industry loss mark is said to be important for many reinsurance programmes and also for a number of ILWs. For catastrophe bonds it’s a little more tricky to assess what transactions may and may not be at risk. A $30 billion industry loss would put much more of the final bill for Sandy in the hands of reinsurers, including a number of collateralized players who we understand have some quite significant exposure in the region through private transactions. For cat bonds, the $30 billion mark could begin to see indemnity losses come into play and the aggregate bonds would likely take on much greater dents to their aggregate layers of protection.
To try to better understand the potential exposure that catastrophe bonds could face, as the insurance industry loss rises, it’s useful to look at some of the primary insurer sponsors and their market shares in the region affected.
For example, take Travelers who are said to be the biggest market player in the region with almost 9% market share. What if they took 9% of the industry loss? That could be over $2 billion and guess what the indemnity trigger for Long Point Re III Ltd. is? You guessed it, $2 billion. Now that cat bond has a clever clause in it which sets a maximum loss per risk, or building covered under the transaction, of $20m. Clauses like tha which are cleverly written into a cat bond can be the difference between no loss and a complete loss of principal. Travelers other cat bond, Longpoint Re II Ltd., which uses an industry loss index and state based payout factors, is harder to assess for riskiness but given Travelers market share in the affected region if the industry loss approached $30 billion we believe this bond will become a concern for investors.
Next consider Liberty Mutual, who have somewhere around 7% to 8% of the market in the north-eastern U.S. states. They have the Mystic Re III Ltd. cat bond in play currently which is an indemnity triggered bond. The Class B notes attach at just $1.3 billion, while the Class A notes attach at $2.1 billion. If Liberty Mutual were to really take 7% of even a $25 billion industry loss that would equate to a $1.75 billion indemnity loss to Liberty Mutual. Make it a $30 billion industry loss and the Class A tranche look risky.
What of Chubb, who have the East Lane Re IV Ltd. cat bond which is one that some investors we’ve spoken to have some concern about. Chubb has around 5% market share in the region, but also have more exposure than most insurers to flood through their commercial programs. East Lane Re IV is an indemnity deal with an attachment on the Class B notes of $2.15 billion. If Chubb took 5% of the losses from a $25 billion industry loss then the cat bond should be safe, make that a $30 billion loss and it’s looking a little more risky, now factor in a greater share of commercial business and it’s easy to see why this cat bond has priced down so much since Sandy.
Another bond sponsored by primary insurer Assurant, Ibis Re Ltd. (Series 2010-1), keeps coming up in conversation as one to keep an eye on as loss estimates rise. This uses an industry loss index trigger but we’re told that in an investment managers internal modelling the Class B tranche of notes look potentially at risk for a $25 billion plus industry loss. Similarly the Class B tranche of the more recent Ibis Re II Ltd. (Series 2012-1), while unlikely to be triggered, may get quite close to attachment if industry losses rise to the $30 billion mark.
We hear that Flagstone Re’s Montana Re Ltd. (Series 2010-1) is another cat bond whose fate is uncertain as industry losses rise. This deal uses an RMS Paradex trigger, which is a unique feature but also relatively untested to date, and the Class D tranche of notes begins to look risky in some analysis we’ve seen as losses rise.
Then of course there is Swiss Re’s Successor Class V – F4 bonds which are considered one of the most at risk. These notes are triggered using an industry loss index so again a rising industry loss puts them more at risk. Swiss Re have another bond which is considered potentially exposed to Sandy, the USW (U.S. wind) tranche of Globecat Ltd. which uses state-based industry payout factors we believe.
We must also consider the aggregate cat bonds, such as the Residential Reinsurance 2011 Ltd. (Series 2011-1) and Residential Reinsurance 2012 Ltd. (Series 2012-1) transactions which are both on ratings watch and expected to accumulate more qualifying losses, and also Combine Re Ltd. (Series 2012-1). While these are thought to currently be safe from any principal losses, a rising industry loss will no doubt result in greater erosion of their aggregate layers, thus making them more risky for the remainder of their risk periods and closer to their trigger points.
Finally, we want to briefly discuss a range of cat bonds that haven’t yet been mentioned as potentially at risk. Allianz has a series of cat bonds with exposure to northeast U.S. wind but they are all triggered on a modelled industry loss basis. This makes it incredibly hard to forecast whether they are at risk or not, however we have seen some internal modelling from an investment manager which suggests that Blue Danube Ltd. (Series 2012-1) Class A notes, Blue Fin Ltd. (Series 3) Class A and B notes and Blue Fin Ltd. (Series 4) Class B notes may all get quite close to attachment points and even have the potential to exceed them at high industry losses. As we said though, it’s extremely difficult to assess these bonds potential exposure or otherwise without access to the models which would run the modelled loss calculation and the underlying exposure data, so this is really just supposition at this time.
It’s worth noting that the loss estimate that will really matter for industry loss based catastrophe bonds is the one that PCS will report at some point after the event. That is the figure used for calculation of attachment or otherwise for many of these bonds. The PCS estimate even two years after Katrina was only just over $41 billion, so much lower than the generally accepted $71 billion that Katrina is now thought to have caused. So while Sandy’s losses could increase by a similar factor to Katrina’s over time, the PCS number was actually only around 140% higher than the preliminary risk modeller estimates. In the case of Sandy, if we scale up the early mid-point estimate of $7.5 billion by 140% we only reach $18 billion. Again, this throws another factor into play that needs to be considered when thinking about cat bonds at risk.
There are other cat bonds which would be considered at risk should insured losses from Sandy leap dramatically, but at this point we don’t feel that is a likely scenario. With all the cat bonds we’ve mentioned above, while industry losses remain below $20 billion the impact is expected to be low to zero, as we discussed in this article yesterday, it’s only once industry losses start to creep upwards further that more and more of these cat bonds come into play and look risky.
We must note that the above calculations and assessment of which cat bonds are at risk is rudimentary and the likelihood of any one insurer taking on losses at an exact relationship to their market-share is slim. Although conversely, there is an equal chance that one insurer could take on losses at a greater rate than their market-share would have you believe. It’s important to note that at this time and if losses don’t rise further than $20 billion many observers have said that they don’t expect any loss to any of the exposed catastrophe bonds. So please treat these insights as something for discussion. Feel free to tell us your opinion below, we’re very interested to hear any insight you have that can be shared.