For mortality linked catastrophe bonds, such as AXA Global Life’s €285m Benu Capital Limited deal that has recently completed, uncertainty as well as the potential for unmodelled risk impacts are always a factor.
Covering extreme mortality catastrophe events is an uncertain business, given there are so many diverse factors that can affect the mortality experience of a sponsor, both positively and negatively. The ceding insurers want coverage for spikes in their mortality experience, above reference mortality index levels, hence the transactions are often called excess mortality covers, the excess being above the index trigger point.
Benu Capital is a five-year deal which is sponsored by AXA Global Life as the insurer seeks out some capital markets sourced collateralized reinsurance coverage for extreme mortality losses. In the pre-sale report published by rating agency Standard & Poor’s a number of the uncertainties surrounding these mortality cat bond transactions are pointed out.
S&P highlighted the risk that unmodelled perils can pose to such a transaction. Very large events, such as catastrophes, occurrence of a war or major terror attacks, could potentially cause such huge loss of life that a mortality index could rise significantly. Even if such an event did not trigger a mortality bond instantly, it could erode the aggregate index level to such a degree that the risk profile rises significantly.
Unmodelled events that could potentially trigger the notes include tsunamis, epidemics, perhaps war or a major terrorism event. Catastrophe bond investors may also consider such perils as a meteorite impact or volcanic eruption, especially now that we find these unmodelled perils in catastrophe exposed bonds as well.
Of course, it’s impossible to model all the potential events that could impact a mortality transaction, meaning that some unmodelled risk will always be present. However it is important for investors to consider the risks posed and to ensure that they are comfortable with assuming a level of unmodelled risk uncertainty.
Similarly there is uncertainty in mortality cat bonds due to medical advances and changes in lifestyle, which can affect mortality rates dramatically. Scientific development of vaccines and other drugs could reduce mortality rates dramatically in the future. However changes in lifestyle could affect them negatively as well.
This uncertainty could be positive or negative to a bearer of mortality risk. Of course some bearers of mortality risk also hold longevity exposures, through other strands of their business, making this type of uncertainty potentially volatile.
Also a minor concern on mortality cat bonds, noted by S&P, is the fact that as world populations grow the frequency and severity of future pandemics or epidemics can increase as well, due to factors such as increasing population density and rising global air travel rates.
In the future the outbreak of a pandemic, epidemic or previously unknown disease, could have a tremendous impact on mortality rates, especially in our more connected and populated world.
Of course these extreme tail risks and the uncertainty associated with them is mitigated, as much as possible, through stress testing and scenario risk modelling.
S&P explained that in order to counter these factors the risk modelling for the Benu Capital transaction stresses pandemic and epidemic risks, and also reviews and stresses mortality improvement factors as well. S&P itself also considered a qualitative assessment of various event scenarios that could affect the bond and trigger any default.
In terms of the main exposures the Benu Capital mortality ILS faces, the risk models used to assess the risk of default, by modelling firm RMS, were:
- The base-case model reflects the expected mortality, including improvements, and variations in mortality during normal times, that is, when no excess mortality event has occurred.
- The RMS infectious disease model reflects additional mortality that could result from the outbreak of certain infectious diseases, including the risk of pandemic influenza and emerging infectious diseases.
- The RMS earthquake model reflects additional mortality that could result from severe earthquake events in the U.S. and Japan.
- The RMS probabilistic terrorism models for the covered area reflect additional mortality that could result from the occurrence of severe nonmilitary acts of violence in U.S. and France.
- The residual model attempts to quantify potential losses from perils (for example, floods, extreme temperatures, accidents, mass movements, wildfires, and volcanic eruptions). It is a statistical, rather than an event-based, model.
S&P explained that although other factors could cause an increase in mortality risk, as explained already in this article, RMS does not consider them as major contributors to the mortality risk in this deal, and hence not critical for assessing the probability of default.
A big proportion of the risk with this deal is tied to the infectious disease modelling, demonstrating where the coverage is really focused for AXA Global Life.
In terms of scenario modelling of known events, RMS found that a recurrence of the 1918 flu pandemic caused by a virus would result in 100% losses on both the class A and B notes. However, if the outbreak was bacterial in nature only a partial loss of about 18% would be reported on the class A notes, although a total loss would occur on the class B notes. S&P notes that the scientific community are divided as to whether the event was caused by a virus or bacteria.
Other historical events such as the peak of AIDS deaths in 1987 and the terrorist attacks of Sept. 11, 2001, would not have triggered a loss under either of the tranches of notes, S&P explained.
In S&P’s assessment the biggest risks to the transaction are a man-made catastrophe, such as a nuclear, chemical, or biological war/terrorist attack; a natural catastrophe; or a substantial pandemic with very limited vaccine or known treatment.
The assessment by S&P is useful, as it provides another opinion on the uncertainty and exposures that a catastrophe bond faces. In the case of a mortality cat bond the uncertainty can be large and there are many unknowns associated with the underlying exposures.
However the unknowns and unmodelled risk factors are largely tied to tail events, which are extremely risk remote, so do not affect the overall rating or expected loss metrics for the bond. The important matter is to consider and assess these unknowns and any unmodelled risks in the context of an investor or ILS fund managers risk appetite.
Of course, it should be remembered, that insurance and reinsurance are games of uncertainty. That’s what risk bearers are being paid for, to a degree. What’s important is to be as informed as possible, in order to minimise it.