Understanding Non-Modeled Risk in Cat Bonds

March 25, 2014

The atmosphere is always electric at the annual SIFMA-hosted Insurance- & Risk-Linked Securities Conference. A million handshakes, tons of one-on-one meeting staking place wherever folks can find a private corner, and an array of panels offering innovative thought from true market leaders and advocates.

At this year's conference, held earlier this month in New York City, one panel in particular really resonated with me as a catastrophe modeler-a discussion of how to facilitate insurance-linked securities (ILS) market growth for difficult to model and currently non-modeled risks. As noted by the panel, there are more and more offerings covering risks for which no detailed catastrophe model currently exists. The majority of these risks fall into two categories:

  • Regional perils-e.g., volcanic eruption, U.S. wildfire (outside of California), U.S. flood (although AIR will be releasing a U.S. inland flood model this summer); 
  • Specialized lines of business-e.g., renewable energy assets, aviation, cargo in transit

For an investor to gain comfort with such an offering, the impact of non-modeled risks needs to be assessed-not only for the transaction in isolation but also at a portfolio-level.

Assessing Non-Modeled Risks for a Transaction in Isolation

Let's say an investor is presented with a detailed exposure database to evaluate a collateralized reinsurance opportunity consisting of location details, replacement values, and associated policy terms and conditions for risks in Arizona that are covered for loss from wildfire. Currently, none of the major modeling firms offer a probabilistic wildfire model for states other than California. In lieu of a fully-probabilistic catastrophe model, one way an investor could assess this risk is to perform a geospatial analysis based on the availability of detailed exposure data.

A geospatial analysis can be as simple as identifying which locations exist within relatively hazardous areas. In this example, a relatively hazardous area could be one where significant wildland vegetation, which acts as a fuel for wildfire, is within the vicinity of insured locations. By importing into Touchstone® publicly-available shape files that denote areas with a high vegetation density, an investor could quickly identify covered locations within or near these areas and accumulate the total value at risk of these locations. This accumulation offers the investor valuable insight into the maximum amount of potential loss, which allows them to better evaluate a specific transaction.

Assessing Marginal Impact of Non-Modeled Risks to a Portfolio

But what if your goal as an investor is to see how non-modeled risks covered by a new offering interact with your existing portfolio? You would need a solution that allows you to quantify and organize the risk profile of the new offering in the same perspective as the modeled risks. To perform this marginal impact analysis, you can use the new Exposure Management Module in AIR's CATRADER®.

One of the many features of the module is the ability to create an event catalog for non-modeled lines of business based on available information and the user's expertise with such risks. Using an event catalog-based approach, you can account for both event frequency and severity as they pertain to this specific line of business by leveraging the same catalog framework used for all of AIR's models.

Consider a new catastrophe bond that covers actual losses from wind turbines after a U.S. hurricane. Current catastrophe models do not have a distinct damage function for wind turbine structures. Using the "Custom Line of Business" functionality in the Exposure Management Module, you can create an event loss table containing loss estimates at a geographical resolution as fine as county-level. These loss estimates can then be merged with the loss estimates of other modeled risks covered by the transaction and then evaluated against your existing portfolio. Here's how:

Step 1
Review the wind turbine exposure. Understand where the assets are located and study the historical loss experience for this line of business. Figure out what types of events have caused damage to these wind turbines as well as what other types of events could potentially cause damage.

Step 2
Using what you learned in Step 1, identify stochastic events in AIR's U.S. hurricane model that may cause some level of loss to these structures based on the stochastic event characteristics and industry loss footprints.

Step 3
Determine a wind turbine loss estimate for each selected stochastic event and simply enter them into the event loss table template.

Step 4
Merge loss output with remaining modeled losses covered by the transaction and compare against your existing portfolio to quantify the marginal impact of this transaction.

If the discussions at SIFMA are any indication of where the insurance-linked securities market is heading, then investors will need to find ways to objectively evaluate risks that either cannot currently be modeled or are simply difficult to model. Whether you are evaluating the risks covered by a catastrophe bond in isolation or trying to understand the marginal impact to your existing portfolio, leveraging AIR's flexible tools will put you in the best position to perform the necessary critical analysis in order to truly own your risk.

If you have any thoughts or questions, I would love to hear from you. Happy (un)modeling!

Don't miss a post!

Don't miss a post!
Subscribe via email:


Loading Video...



You’re almost done.
We need to confirm your email address.
To complete the registration process, please click the link in the email we just sent you.

Unable to subscribe at this moment. Please try again after some time. Contact us if the issue persists.

The email address  is already subscribed.