Key takeaways
  • Multi-region catastrophe triggers
  • Loss dependence shapes price
  • Fixed and random loss amounts
  • Property Claim Services fit

When the same disaster can strike more than one region, the price of protection shifts in ways that simple models miss. This paper studies a new kind of contingent convertible bond, or CoCoCat bond, whose trigger is tied to predefined natural catastrophes across multiple geographical regions. The authors test several ways those regional losses can relate to one another, from complete independence to proportional loss dependence, and they examine both fixed and random loss amounts. Using change-of-measure techniques, they derive risk-neutral pricing formulas for each setup. They then fit the model to real natural catastrophe data from Property Claim Services and show that inter-regional dependence has a significant impact on the bond’s price. The message is straightforward: if you are pricing catastrophe-linked protection, you cannot treat each region as if it lives in its own separate world.

Imagine a bond that pays off unless disasters hit more than one region. That sounds tidy on paper. But the price turns on a hidden question. Do regional losses act like strangers, or like neighbors who fall together? This study looks at CoCoCat bonds, short for contingent convertible catastrophe bonds. They change shape when a preset disaster trigger fires. The surprise is simple. The same bond can cost very different amounts once regional losses move together. A buyer who ignores that link may pay for the wrong risk. A seller who sees it can price the bond with more care. That is the gap this model tries to close. It treats a joint storm map as part of the deal.

When one storm is not just one storm

The model tests two ways regions can line up. One treats losses as independent. The other ties them together through proportional losses. Each setup also comes with fixed and random loss amounts. That makes four useful cases to compare. The result is not subtle. Inter-regional links can move the bond price a lot. The abstract says these links have a significant impact on pricing. So a multi-region trigger is not just a bigger version of a single-region one. Its value depends on the shape of the joint risk. Two regions can look calm on their own. Together, they can still push the bond into a new price zone. That is the core message from the fit to real loss data.

2dependence structures

independence vs proportional loss

regional loss models
  1. One case keeps regional losses independent.
  2. Another case ties them together through proportional losses.
  3. Each case can use fixed loss amounts.
  4. Each case can also use random loss amounts.

How the price gets built

To price the bond, the model needs a fair-odds lens. Finance calls that a risk-neutral price. That means a price built from market odds, not fear. This model uses change-of-measure techniques, a math swap that moves from real-world loss chances to that fair-price world. It then writes formulas for each dependence case. Fixed losses make the math cleaner. Random losses let payouts vary more like real disasters. The point is not a trick for its own sake. The point is a price rule that can handle several regions at once. That matters when one trigger can watch more than one place.

“the significant impact of inter-regional dependencies on the CoCoCat bond’s pricing”

From the abstract

Why the result matters

For sponsors, this changes the whole price talk. A sponsor is the side that buys protection. If regions do not move the same way, the bond is not one neat bet. It is a web of linked bets. That matters for insurers, reinsurers, governments, and companies that use insurance-linked securities, or bonds that pass disaster risk to investors. These tools move catastrophe risk from the sponsor to investors in the wider market. The model shows that pricing should not ignore how one area's losses line up with another's. If it does, the price can miss the true risk. The fit to Property Claim Services data backs that warning. A single-region lens is too blunt for a multi-region trigger.

What to watch next

The surprise is still the same. Regional links can move the price on their own. That means pricing desks cannot treat regional dependence as decoration. In a multi-region CoCoCat bond, the joint pattern of losses is part of the price itself. The fit to Property Claim Services data backs that up. So the habit of pricing each place on its own can miss the mark. For anyone designing cover against hurricanes, quakes, or other cat losses, the message is blunt. The map of how regions move together belongs on the pricing sheet. That is the concrete shift this model makes.