The complete loss of the first U.S. tornado catastrophe bond less than a year after its launch is raising questions among investors about how, or whether, insurance-linked securities can back the risk. 
At issue is how the bond — Mariah Re — slowly built up losses then suddenly became a complete loss for investors in the $100 million note.
“We were extremely surprised by the loss development,” says Dirk Schmelzer, portfolio manager for ILS products at Plenum Investment in Zurich and an investor in Mariah. “Not that the bond could go to zero, but how it went to zero.”
Mariah Re was issued in November of 2010 with great fanfare as the first catastrophe bond to move away from large perils in North America, such as major hurricanes and earthquakes, to cover a smaller but more frequent storms like tornados and severe thunderstorm. The bond’s sponsor is American Family Insurance of Madison, Wisconsin.
The key for the bond in covering smaller storms was its “aggregation” feature that applied different weighting methods to measure losses that would lead to the trigger. The structure would rely on loss reports from ISO’s Property Claim Services for measure losses and then, based on those reports, apply the weighted losses to the bond until it was triggered.
Although unique and somewhat complicated, the bond peaked investor interest and specialist firms like Plenum snapped up issuance while others were less enthusiastic.
“When it came to the market is was a novelty,” says Michael Stahel, head of insurance linked investments for Clariden Leu AG in Zurich. “It was a first of its kind and the modeling for tornado risk was still fairly new – so you need to apply proprietary modelings as an investor in this space.”
Mariah’s fortunes quickly changed, however, when a series of severe storms and tornados in the U.S. in April and May of 2011 quickly mounted losses. An EF 5 tornado alone struck the city of Joplin, Missouri on May 22nd causing an estimated $2.2 billion inured losses alone.
Investors were put on notice, but reports from PCS after the storms showed that while losses mounted to $825 million attachment point — and growing — they were unlikely to reach the bonds exhaustion point of $925 million.
Investors, like Plenum, backed off from selling off the bonds for a loss in the secondary market thinking they could still retain some value.
“Our decisions making whether to sell the position in the summer, when the first losses were taken, were framed out of the event reports by PCS,” Schmelzer says.
But last month PCS updated the report on the storms, classified as “Catastrophe 42” by $118 million to $954.6 million, according to ratings agency Standard & Poor’s. The bonds are now considered a “100% loss” , S&P said, and investors were essentially wiped out.
“We were surprised to find that PCS has the ability to change information going backwards,” says Schmelzer. “It is strange that, if you are reporting agency, you can change investor relevant documents without make any notice.”
The key to Plenum’s frustration was not only that PCS updated loss amounts, but how. Schmelzer argues that the last report issued in November changed several of the events from “non-metro” into “metro” events, a change aimed directly at the terms of Mariah’s unique losses weighting system. “Did they try and sneak in a list of cities? Did they forget them? It comes down to,I suppose, do you do American Family a favor or do the cat bond investor a favor,” he says.
Representatives from PCS and American Family did not respond to several attempts for comment.
Some signals of Mariah’s possible issues with its aggregation and weighting methods were explicit, Stahel argues, and could have been detected with proper due diligence.
“Based on our assessment and – without applying an inflation rate – one of the last 10 years on record would have been close to attaching to the first layer in terms of actual tornado activity and losses – and by applying a modest inflation rate, that year would have leaped into the first layer.” Stahel explains. “It is certainly not the case that the modeling was wrong, but it pretty much comes down to the actual geographical spread and weighting of the different losses.”
Stahel dismisses the idea that investors could have been taken off guard by a full loss on Mariah Re and that the aggregated loss structure was a known downside.
“One of the reasons we did not participate on the transaction was the aggregated loss structure, because it means the more events you have during the year the more you would see a mark-to-market impact on the bond even though there was no actual payout yet. And there is a certain time-lag in reporting – but this was all transparent in the offering and it applies to all aggregate transactions.”
“No investor that invests in Mariah Re should be surprised to loose money after such a horrible Tornado season – after all, this is big boys business and that is the risk investors take,” Stahel argues.
Schmelzer, however, says that his firm’s experience with Mariah will make him hesitate from investing again on U.S. tornado risk.
“The market will look at the peril of tornado risk differently in terms of pricing, modeling and aggregate structures and its dangers differently is dangers,” he says. “There has to be clear definition of how PCS reports. That has to be fixed.”
Ironically, Stahel says, he would be inclined to seriously consider another tornado bond if it come to market even thought he took a pass on Mariah.
“We will now potentially see more such tornado transactions, some with higher attachment levels and I guess the modeling will imply high probability of loss after such a year – but the deals will then also pay more,” Stahel says. “Now it may be time for us to step in.”
