by Pascal Karsenti, CCM, Manager, Insurance Linked Securities, AIR Worldwide
In today’s burgeoning insurance-linked securities (ILS) market, investors are increasingly using catastrophe modeling software to “remodel” catastrophe bonds. Remodeling—which differs from the initial modeling process, as described in Table 1—helps investors gain additional insights into a bond, such as its impact on their existing portfolio. It also serves those who issue the bond by offering a process through which increasingly sophisticated investors can gain additional confidence in a bond’s risk characteristics, thereby facilitating their decision to invest.
To commence remodeling, investors must possess information on a bond’s structure and underlying exposure. This information can be found in the bond’s offering materials, which those involved in bond issuance spend substantial time and effort to create. Clear and complete disclosure in the offering materials—again, particularly of the exposure—is critical to ensure the highest quality remodeling possible.
Today’s investors are able to leverage the information disclosed in a bond’s offering documentation through catastrophe bond remodeling services, which AIR pioneered more than ten years ago and which are explained in greater detail below.
Pascal Karsenti
Investors remodel cat bonds for numerous reason, including: to obtain customized cuts of the risk; to assess the impact of potential or actual catastrophic events on a portfolio of bonds; to better understand risk correlations and diversification effects; to understand the impact on their risk profile of changes in catastrophe models as they evolve over time; and to better understand the uncertainty in the risk estimate of a particular bond.
HOW DOES A CAT BOND REMODELING SERVICE OPERATE?
The remodeling process is initiated when an investor requests an analysis for a particular transaction and then provides the relevant offering materials to their catastrophe (re)modeling company of choice. The company enters the exposure information and contract terms as delineated in the bond’s offering documentation into a catastrophe modeling software platform and generates a detailed report showing the risk profile of the transaction.
In addition to the report, the company also provides an input file to the investor to perform their own risk analyses in the software; because such files are available for all outstanding catastrophe bonds (and many industry loss warranties), investors can evaluate their complete portfolio of ILS holdings in one platform.
Assessing the overall portfolio risk within a single platform is crucial for investors because their investment guidelines—the conditions under which they can invest their funds—will typically require that they respect certain risk accumulation metrics for given regions and perils. Without the ability to analyze all cat bonds using a single platform against the same set of simulated events, investors may resort to controlling their accumulations simply by adding up the total value of their bond investments exposed to a given area or peril, regardless of the riskiness of such investments. This limits the total capital that investors can commit to a given peril. The availability of high quality bond remodeling thus ultimately helps investors consider more bonds exposed to a given peril, while still meeting their investment guideline.
HIGH QUALITY OFFERING MATERIALS FACILITATE THE PROCESS
The outcome of the remodeling process is impacted by three primary factors: the bond’s structure and trigger type; the exposure and structure information disclosed in the offering documentation; and differences in the catastrophe models used for the initial risk analysis and remodeling.
This section will review best practices for the disclosure of exposure information in offering documentation, which greatly facilitates the remodeling process. These best practices depend on the bond’s trigger type.
• INDEMNITY bonds trigger based on actual losses to the cedant’s exposure. Remodeling quality for indemnity triggers is dependent on the availability of exposure information at a high resolution; for example, by county in the U.S. and by CRESTA zone in Europe.
When an indemnity bond covers multiple perils, such as earthquake and tropical cyclone, best practice is to disclose separately the exposure for each peril. If the bond is exposed to different portfolios for sub-perils, such as a small portfolio of earthquake shake exposure and a larger portfolio of earthquake fire following exposure, then it is in the interest of the cedant to disclose the individual portfolios lest the modeler assume that both sub-perils (in this case, shake and fire-following) are exposed to the larger portfolio. Such disclosure techniques increase remodeling quality, as does the breakdown of exposure by line of business.
Presenting a special challenge in the realm of the indemnity bond is the securitization of portfolios of commercial properties where the insurance company only insures a fraction of the risk for each asset or offers surplus share protection. For such portfolios, modeling the total sums insured on a ground-up basis could overstate the losses, while modeling only the total limits would mean losses were understated. Cedants can help modelers and investors interpret such exposure by providing ratios of gross to net losses; such ratios, once applied to the sums insured, approximate the impact of the partial coverage.
When it comes to bonds made of a portfolio of reinsurance contracts, properly modeling the risk requires that the exposure be separately disclosed for each of the reinsurance contracts included in the bond. While this may be possible for a relatively small portfolio of reinsurance contracts, it is not practical for larger portfolios consisting of hundreds of treaties. In such situations, providing an indication of the amount of industry loss likely to cause a loss to the bond will allow the remodeler to perform a proxy remodeling based on industry exposure and loss. It will also increase the bond’s transparency.
• INDUSTRY LOSS bonds trigger based on actual losses to the insurance industry as a whole in a specified area, as reported by a third party such as Property Claim Services, the Verisk Catastrophe Index, or PERILS AG in Europe. These bonds are often exposed to a fraction of the industry loss determined by payout factors that may vary by state, county or CRESTA. Industry loss triggers can be remodeled as long as both the payout factors and the layer of the bond are disclosed. This allows remodeling firms to provide results of a quality similar to that which would have resulted had they had performed the initial risk analysis themselves.
• FIRST-GENERATION PARAMETRIC bonds trigger based on the occurrence of an event with specified parameters, such as an earthquake above a given magnitude or a hurricane above a specific Saffir-Simpson category occurring within a given area. Because the structure is typically fully disclosed, including earthquake magnitude and depth requirements, or hurricane intensity requirements, these bonds are straightforward to remodel using the information in the bond’s offering documentation.
Beyond disclosure of terms, best practices for such transactions are to ensure that investors can easily estimate potential payouts for themselves following an event. This can be done by triggering the transaction based on event parameters easily accessible to investors through public reporting agencies’ websites, such as that of the United States Geological Survey or National Hurricane Center. Cedants can also provide a direct link to this data in the offering materials, and even include a spreadsheet allowing investors to input event parameters and easily estimate whether the bond may suffer a loss for a given event.
• SECOND-GENERATION PARAMETRIC bonds trigger based on the wind speed or ground motion detected by a network of localized recording stations, and they generally require the application of formulas to determine whether a bond will suffer a loss. These bonds can be complex, and care must be taken in the interpretation of the parameters required to estimate their risk profile. This complexity can make second generation parametric indices highly sensitive to uncertainties in measurement.
Clear and straightforward disclosure of the required parameters is critical for these transactions, as any uncertainty linked with the required parameters can lead to more uncertainty in remodeling the bond. Further, the more complex a parametric structure is, the more small differences in the science or methodologies of two firms modeling the same bond may be amplified, potentially causing the two firms’ estimates of risk to diverge.
• MODELED LOSS bonds trigger not on actual losses to a cedant’s portfolio, but on the modeled losses from an actual event, based on their underlying exposure. The risk analyses for modeled loss are ostensibly similar to those of the aforementioned indemnity transactions. Companies may view it as an advantage that, for modeled loss bonds, the cedant may not have to disclose actual exposure, but can instead disclose disguised or proxy exposure—as long as that is the exposure against which the bond will trigger. Thus, for this trigger type, cedants may disclose significant details about the bond without necessarily revealing their own, confidential exposure data.
Today’s catastrophe bond investors have access to powerful remodeling services offered by the leading catastrophe modeling firms, including AIR Worldwide. These services perform a valuable role in the ILS market by allowing increasing numbers of investors to perform in-depth analyses based on a bond’s offering documentation. Because remodeling services are based only on information disclosed in bond offering materials, they are subject to important limitations: namely, the quality and quantity of information disclosed to investors about the structure of the bond and the exposure at risk. As this information becomes more detailed and complete, investors are better equipped to make critical investment decisions. By serving investors in this way, remodeling services such as AIR’s Reanalysis Service are contributing to the growth and expansion of the ILS marketplace.
AIR’s remodeling service, known as its Reanalysis Service, is based on the CATRADER® platform, the same platform used by 90% of reinsurers to assess their exposure to catastrophe risk. It is also the same platform used by dedicated ILS investors and reinsurers to manage their cat bond portfolios.
Table1: Some key differences between the initial risk analysis and remodeling
REMODELING |
Commissioned by and performed on behalf of the issuer. |
Commissioned by and performed on behalf of investors. |
Focuses exclusively on the results of a single model, | Compares and contrasts results between different models’ results. |
Performed in advance of issuance. | No advance notice. Must be performed quickly, preferably during the bond’s marketing period. |
May not be updated over time. | Always uses the most recently released model. Model changes means that there could be substantial divergences of view between the risk analysis and the remodeling, even when both were performed by the same firm. |
Performed with full access to detailed information on the cedant’s portfolio. | Performed based on the information available to investors in the bond’s offering materials. |
Pascal Karsenti is a manager within AIR’s Consulting and Client Services group. He is responsible for the oversight of AIR’s capital markets activities, and has been involved in the issuance of over 30 catastrophe bonds.