Japan: Catastrophe bonds dodge quake payout

	"Earthquake triggers on most cat bonds with Japan exposure tend to be measured in terms of geographic position and ground acceleration, usually severe ground motion in or around Tokyo. If the event does not meet these parameters as defined by the reinsurance contract, investors face no losses" Sandro Kriesch, Twelve Capital

The costliest natural disaster of all time could drive further growth for the $12.9 billion ­catastrophe bond market, according to specialist investors who expect greater awareness of cat bond risk-adjusted returns to attract new capital into the sector.

"Landmark events like Hurricanes Katrina and Andrew in the US caused capital to flow into reinsurance and acted as a development catalyst for the cat bond market. If losses from the Japanese earthquake and tsunami cause premiums to rise, it could again widen the investor base," says Sandro Kriesch, partner at Twelve Capital, a Zurich-based investment fund specializing in natural peril catastrophe risk and the extreme mortality risk posed by pandemics.

At less than 1% of the Japanese government’s projected $309 billion total economic loss from the earthquake, tsunami and nuclear crisis, the $1 billion of catastrophe bonds potentially facing principal losses seems confusingly low. Oakland, California-based catastrophe modeller Eqecat explains that although total insured losses could range from $15 billion to $22 billion, global reinsurers’ exposure will be capped by Japan’s earthquake reinsurance scheme, which is expected to foot $2 billion to $4 billion of the non-life losses.

Active reinsurance buyers

"The range of insured losses affecting Japan’s kyosai (commercial/industrial insurers) is around $4 billion to $7.5 billion. This exposure is significant for the global reinsurance providers because kyosai are very active buyers of reinsurance cover," says Tom Larsen, senior vice-president and product architect at Eqecat.

Cat bonds are fixed-income securities that insure the sponsor of the bond against natural disasters by transferring a precisely defined set of risks to capital markets investors. For insurers, they typically form one layer in a stack of protection measures, allowing them to shift the most remote and extreme event risks off the balance sheet, freeing up capital to write new policies. The insurance contracts underlying the structure typically specify the geographical regions and types of natural disaster covered by the transaction. Crucially, this agreement sets the severity conditions that will trigger investor payments to cover insured losses.

Investors say that although it is too early to know the exact level of triggered payments, the seismic pattern of the Tohuku earthquake could fall outside the conditions set by most cat bonds with Japanese earthquake exposure.

"Earthquake triggers on most cat bonds with Japan exposure tend to be measured in terms of geographic position and ground acceleration, usually severe ground motion in or around ­Tokyo," says Kriesch. "If the event does not meet these parameters as defined by the reinsurance contract, investors face no losses."

Although US scientists said that the earthquake moved the island of Honshu 2.4 metres east and shifted the Earth on its axis by 10 centimetres, Eqecat’s Larsen says that the pattern of seismic activity in Japan did not meet the event description contemplated by most of these "parametric" catastrophe bonds. "Where most Japan-linked parametric deals are triggered by ultra-severe ground motions in a distinct geographic area, the recent event was modestly severe compared with the triggers and spread over a very large area," he says. Given the scenes of devastation following the earthquake it seems inconceivable that the ground motions might not meet event description criteria – and might prompt some issuers to question how effective cat bonds are at protecting them against this type of risk.

Difficult to quantify

John DeCaro, founding principal and portfolio manager at Elementum Advisors, a New York-based asset manager specializing in securities linked to natural event risk, says that a clearer picture should emerge in April, when some of the transactions under scrutiny report to investors. "The reality of this is that the post-event impact of earthquake risk covered by catastrophe bonds is much more difficult to quantify than, for example, hurricane risk," he says. "There are several standard structures, all of which require several different types of information to determine whether triggers have been activated. Typically, the range of information transparency for investors is very wide and varies by transaction."

Moody’s credit rating agency last month placed four transactions on negative credit watch with a combined outstanding balance of about $570 million including one deal traded by Munich Re on behalf of kyosai Zenkyoren in 2008.

"Japan earthquake is the only peril covered in the transaction... qualifying events covered include any earthquake event occurring during the Risk Period in Japan, and for which Peak Ground Acceleration is at least 0.05 times the acceleration of gravity in at least one location," the agency said.

Meanwhile, AM Best, a New Jersey-based rating agency, placed $200 million of double-B rated notes sponsored by Platinum Underwriters Bermuda under review after the issuer requested that the "calculation determines whether an activation event or loss event has occurred with respect to the Japan earthquake of March 11, 2011". The deal in question provides the issuer with second and subsequent event coverage for a range of disasters including Japan earthquakes.

Market reaction

Secondary market reaction has ranged from moderate to extreme. Deals perceived to have a high likelihood of being triggered by the Japan disaster have lost as much as 60% in face value, investors say. "The market appears to be reaching consensus that there will be a loss of principal to at least one deal but until all the data is collected, the exact amount is currently unknown. Bonds exposed to Japanese earthquake risk are showing stressed secondary market levels ranging from the mid 90s down to the high 40s," says DeCaro, depending on the type of trigger mechanism. But little or no trading has occurred in the most impaired names. Data compiled by Swiss Re Capital Markets point to similar short-term weakness, the Swiss Re Cat Bond Performance Total Return Index losing 1.7 points from 97.39 to 95.69.

Last month, European reinsurer Munich Re gave guidance to shareholders that €1.5 billion of new payment claims for the Japanese earthquake and tsunami might force it into loss in 2011, after previously forecasting an expected €2.5 billion profit. The reinsurer said the unusually high concentration of natural disasters in the first quarter of the year, including the earthquake in New Zealand, and floods and cyclones in Australia, left it with little capacity to absorb the Japanese disaster. Munich Re’s share price fell more than 6% after the event.

"The key uncertainty in cat bond default probabilities is related to the science of earthquakes. There is a substantial body of scientific evidence which forms the basis of the calculation. A parametric hurricane bond, for example is triggered only by wind speed" Tom Larsen, Eqecat

With the usually expensive US hurricane season starting on June 1, institutional reinsurance providers say the shortfall between provision and actual insured losses is driving an increase in new coverage agreements, which could now cost more. "Some European reinsurers say that they have already exhausted their protection as of end of Q1," says Twelve’s Kriesch. "With more companies looking for additional protection, the reinsurance market is definitely not going down. We expect reinsurance pricing to increase in specific areas like Australia and Japan."

While the lack of credit rating on many cat bond deals has deterred broader participation from the institutional investor base, the central role of actual historical catastrophe data in setting cat bond loss probabilities could provide greater comfort than traditional rating agency analysis. "The key uncertainty in cat bond default probabilities is related to the science of earthquakes," says Eqecat’s Larsen. "There is a substantial body of scientific evidence which forms the basis of the calculation. A parametric hurricane bond, for example is triggered only by wind speed."

Kriesch concurs. "Cat bonds are typically modelled to a much higher level than a comparable corporate bond, where all the risks are modelled on a stochastic, or non-deterministic basis," he says. "In general, cat bonds are one notch too conservatively rated." Although yields have come down since 2009, cat bonds typically generate returns in the Libor plus 100bp region, generally comparable to high-yield corporate bonds and other sub-investment-grade fixed-income sectors. Yields can be lower for more diversified risk exposures, Kriesch says.

Although taking exposure to natural catastrophe risk is a specialized, scientific undertaking and will probably not appeal to traditional fund managers, the default rate might provide food for thought. In the market’s 12 years there has been just one catastrophe-related default (ignoring the technical default of four deals underwritten by Lehman in 2008) when Hurricane Katrina triggered a $144 million payment from holders of Zurich Re’s $190 million Kamp Re 2005 transaction. Investors familiar with the deal say that it redeemed some at around a quarter of face value.

Extreme events do happen

Thanks to Japan’s national insurance subsidy, it seems that reinsurers have largely escaped financial responsibility for the most costly disaster of all time. However, the doomsday prospect for the reinsurance industry, a magnitude seven or higher earthquake in either California or Florida, remains at the forefront of investors’ minds. As Tohuku shows, however, unexpectedly extreme events do happen, and usually serve to recalibrate expectations about what is possible. "The Tohoku quake came from seismic forces that were not previously thought to have the potential to cause the amount of damage that they did," says DeCaro. "Not in their worst case scenario analysis did Japanese seismologists expect four of the five segments of the Japan trench rupturing at once."

Eqecat is expected to improve on the science. In Japan, for example, its model predicted an extreme earthquake an order of magnitude 0.25 higher than government scientists, Larsen says: "In the Pacific Northwest, our model suggests that an extreme earthquake event could be as strong as 9.2, which is 0.7 higher than the magnitude currently modelled by the US Geological Service. The associated insured loss could be around $25 billion. The probability that realized events will be much more severe than predicted is certainly a concern, and something that we will be reviewing."