The Hurricane Hunter and the Cat Bond

The second Mexico multi-cat securitization was launched at the end of 2012 to provide reinsurance for the Fund for Natural Disasters (FONDEN) – established by the Mexican federal government as a disaster fund for the poor, which also finances disaster-damaged infrastructure. This three-year bond had a series of tranches covering earthquake or hurricane, each to be triggered by parametric “cat-in-a-box” structures. For one hurricane tranche, this was based on the central pressure of a storm passing into a large “box” drawn around the Pacific coasts of Mexico and Baja California, with the length of the box spanning well over a thousand miles. With a “U.S. National Hurricane Center (NHC) ratified” hurricane central pressure in the box of 920 millibars (mb) or lower there would be a 100 percent payout or $100 million; for a central pressure of 932 mb to 920 mb a 50 percent payout of $50 million.

Hurricane “pressure-in-a-box” trigger structures are notoriously high in basis risk. For earthquake, the stronger the event, the larger the footprint, while for hurricane it tends to be the opposite. Given the degree to which coastal exposure is concentrated in a small number of widely separated coastal cities, such as Acapulco and Cabo San Lucas, the basis risk for this transaction was particularly high, as events would prove.

However, those who designed the parametric central pressure trigger were not aware that its measurement could become hostage to the activities of an amateur hurricane hunter.

“Hardcore Hurricane Chasing”

Josh Morgerman’s obsession with hurricanes began as a boy when in 1985 Hurricane Gloria came and ripped up his parents’ house in Long Island. In 1999, he went on to co-found a successful digital advertising agency in West Hollywood, but it was during the 2004 and 2005 U.S. hurricane seasons that Josh decided to pursue his passion. He invented “hardcore hurricane storm-chasing” under the brand name iCyclone. With iCyclone, Josh’s mission was “to penetrate, experience, and document the inner cores of hurricanes as they thrust ashore — via video, data collection and blogging.” Benefiting from improvements in forecasting, he set out to place himself in a property, ideally a hotel, as close as possible to the hurricane center at landfall. As intense hurricanes stayed away from the U.S., in 2007 he broadened his scope to hunt “ferocious” Hurricane Dean on the Yucatan Peninsula of Mexico, gaining an appetite for Mexican storms.

The tracking of hurricane pressures offshore Mexico’s Pacific coast is made possible through research planes flown down from the U.S. As they fly through the hurricane eyewall, dropsondes are launched to measure low-level winds and pressures. However, for fear of collateral damage, such transects and dropsondes are curtailed over land.

Observations from a Holiday Inn Express

On September 14, 2014, Hurricane Odile made a direct hit on the resort city of Cabo San Lucas in Mexico, located at the southern tip of Baja California. Odile’s final pre-landfall minimum central pressure was 923 mb with winds of 130 knots (kts), with Odile making landfall just to the east of Cabo San Lucas, nine hours later at 04:45 UTC on September 15. The airport weather sensors at Cabo San Lucas International Airport, some four miles (six kilometers) inland from the city, had packed up three hours earlier, so the only landfall pressure observation (at 943.1 mb) was made by Morgerman staying at the city’s badly damaged Holiday Inn Express, located about a mile east of the city, next to a wide stretch of beach.

Based on this observation, the U.S. National Hurricane Center (NHC) estimated the landfall pressure minimum was 941 mb. However, without Morgerman’s observation, the first direct pressure measurement from close to the track of the storm was 12 hours and some 150 miles (240 kilometers) further north at around 973 mb. Without Morgerman’s measurement, reconciling the offshore and onshore pressure observations could have brought the central pressure crossing into the offshore boundary of the parametric box close to 932 mb or lower – potentially triggering a $50 million loss on the bond.

The FONDEN fund was disappointed with the lack of such a payout. Odile was the kind of event for which a trigger would be appropriate, particularly hitting poorer communities and infrastructure. The Mexican insured loss was more than $1.2 billion.  The storm destroyed 550 high voltage transmission towers and 3,400 distribution posts, resulting in around 239,000 people across the Baja California Sur, some 90 percent of the population, being without electricity. The main airport, Cabo San Lucas International, did not fully reopen for a month. Drinking water was unavailable in Cabo San Lucas. Thousands were left homeless, and in many smaller towns along the peninsular people lost everything. In the state capital La Paz, on the east side of the peninsula, some 10,000 citizens were homeless.

Enter Patricia

A year and a month later came Pacific Hurricane Patricia. On October 22, 2015, after some unprecedented intensification, Patricia produced the lowest pressure (872 mb) reading ever observed for a hurricane. A traverse at 18:00 UTC still found surface winds of near 180 kts and a central pressure of 878 mb, but two and a half hours later, in a final traverse before landfall, the central pressure had risen by 24 mb. Patricia made landfall in a low population part of the state of Jalisco, near Playa Cuixmala, around 23:00 UTC on 23 October. Operationally, Patricia was assessed by the NHC to have a landfall pressure of 920 mb which would make it a 100 percent ($100 million) trigger event for the MultiCat bond.

At the time of landfall, Morgerman was in the disintegrating Hotel El Refugio in Emiliano Zapata, a couple of nautical miles inland from the landfall point, where at 23:13 UTC he measured (on two separate recorders) a surprisingly high minimum pressure of 937.8 mb on the eastern edge of the eye. If his was the only observation this would have determined the NHC official landfall pressure, in which case the MultiCat structure would again not have triggered. The bond existed in a state of limbo for the next three months, awaiting the NHC’s final report on the landfall pressure.

Fortuitously, there was another observation, an automated weather station within the private 36,000 acre Chamela-Cuixmala Biosphere Reserve, a couple of miles away from Morgerman’s base. The station was attached to La Loma, a grand, coastal clifftop villa forming part of the ultra-exclusive Cuixmala boutique hotel complex. This luxury villa and estate was owned in the 1980s by the late British billionaire financier Sir James Goldsmith, and is still owned by his family. Although the windspeed recorder failed, the weather station gave a pressure reading of 934.2mb.

On February 4, 2016, the NHC declared a minimum central pressure of 932 mb at landfall unusually acknowledging that there was uncertainty of this value by +/- 2-3 mb. At its most intense, the storm’s radius of maximum winds (Rmax) was only ten kilometers (six miles), which is why the central pressure was able to rise to an overwater record of 54 mb in the five hours before landfall – the circulation was as much a “‘mega-tornado” as a ‘“mini-hurricane”.

Landing in a relatively sparsely populated rural area, even with its Category Four winds, the skinny wind field of Hurricane Patricia was not a big loss event. In terms of the costs to FONDEN this was not a disaster intended to trigger the bond. But after the Odile debacle, it was at least fitting that, based on the NHC landfall central pressure, FONDEN received a partial payment of $50 million for Hurricane Patricia.

This story highlights the basis risk challenges of “pressure-in-a-box” hurricane triggers. The most intense hurricanes (like Patricia) are typically small and can easily miss a concentration of exposure, while a larger but weaker storm (like Odile) can cause significant regional damage.

That the measurements of one amateur hurricane hunter could determine the outcome of a $100 million Mexican catastrophe bond introduces new risks. This situation has the potential to happen anywhere official observations are sparse. What if the hurricane hunter was also an investor in the bond?