Dr Luzi Hitz takes a look under the hood of cat bonds. These bonds employ different trigger mechanisms: rules which determine a pay-out to sponsors.
From a sponsor’s perspective, there are typically three key requirements of the trigger mechanism. Firstly, the trigger must be straightforward, thereby minimising preparation time and structuring costs. Secondly, the trigger must satisfy the transparency requirements of the investor community to avoid any excess premium for trigger uncertainty. And thirdly and most importantly, the cover must ensure a pay-out which is as closely aligned with the sponsor’s indemnity loss as possible.
Trigger conditions in cat bonds are typically grouped into four main categories: indemnity, modelled loss, physical parameter and industry loss. Combinations of these four basic trigger types are common.
About 40% of outstanding US cat bonds are based on industry loss triggers, which are almost exclusively defined by the Property Claims Service (PCS), an independent US-based service provider. However, on the other side of the Atlantic there is not a single cat bond transaction involving an industry loss trigger, primarily due to the lack of independent industry loss data. It is hoped that recent developments in the European market to establish an independent source of insurance loss data will serve to remove this hurdle.
An industry loss is easy to relate to. The challenge for the sponsor is mainly in aligning his own portfolio with the market portfolio. By breaking down the industry loss into various geographical zones with different weighting, this risk factor can however be managed. A parametric trigger allows for closer portfolio tracking, but risk assessment efforts and the associated costs are generally higher.
Parametric transactions provide the investor with the advantage of excluding any unexpected indemnity losses. On the other hand, default probability calculations – and hence pricing – are based on a custom risk assessment with second opinion difficult to come by. In industry loss triggers, the investor is exposed to potentially ‘unexpected’ losses. On the positive side, however, risk assessment for industry losses are standardised and can be tracked over the years.
What really matters in the end to the sponsor is whether they will get their pay-out when the catastrophe inevitably strikes. Parametric triggers are generally faster when it comes to pay-out for the simple reason that it takes considerably less time to run a computer model than to assess an industry indemnity loss. On the other hand, the computer models making the link between parametric trigger and indemnity must be able to account for all factors leading to the indemnity loss. Experience shows that this is a major challenge, not least because of the ‘known unknowns’. Storm surge, fire, dam and levee failure (e.g. Katrina), landslides, soil liquefaction, or long distance tsunamis are all formidable challenges to risk modellers. Similarly, the accurate modelling of complex original cover structures, in particular structured industrial insurance accounts, within the context of an entire insurance company or industry portfolio is another daunting task for even the most experienced modeller. Add to this the ‘unknown unknowns’ and the modelling risk factor is indeed significant. In an industry loss trigger, all these risk factors are not present because the
industry loss is based on the actual incurred loss and includes all loss-contributing factors, whether previously known or unknown.
There is no simple equation for evaluating the most suitable trigger type for a given protection need. In the end, the choice of the appropriate mechanism boils down to an informed management decision. And because every management will have different needs the coexistence of different trigger types will certainly continue.
Dr Luzi Hitz is CEO of Perils AG.
A good trigger mechanism should be:
- easy to set up
- transparent for investors
- able to keep down basis risk