Hurricane Maria
Status: Closed
Post Landfall 4 | Summary
Posting Date: December 6, 2017, 8:15:00 AM
After re-analyzing Hurricane Maria’s meteorology; analyzing findings from AIR’s damage survey conducted in Puerto Rico in late October (which corroborated our new view of the hazard); and reviewing new information about insurance practices and policies in Puerto Rico, AIR Worldwide now estimates that insured losses from Hurricane Maria in the Caribbean will be between USD 27 billion and USD 48 billion. Losses in Puerto Rico are estimated at between USD 25 billion and USD 43 billion. Note that the significant reduction in the high end of our loss estimates is driven primarily by the review of Maria’s intensity over Puerto Rico.
For information on what is and is not included in these loss estimates, please see the final section (below) of this summary.
It is worth noting that a significant portion of the AIR modeled losses for Puerto Rico are generated by the industrial line of business. However, it is also the case that the industrial line accounts for much of the remaining uncertainty in the loss estimates, particularly as it relates to business interruption losses. Adding to the uncertainty (and to the estimated range of losses) is to what extent demand surge, which is the increase in the cost of labor and materials that is often observed in the aftermath of major catastrophes, will play a role in ultimate insurance payouts. In estimating insured industry losses for Maria, we have used our default demand surge function, which contributes roughly 20% to the total loss at the high end of the range.
AIR has posted a discussion paper detailing the revised estimates, the considerations that have informed them, and remaining uncertainties. Hurricane Maria – Understanding AIR’s Revised Modeled Loss Estimates is available (login required) on the Downloads tab of this website. We urge all clients to read the paper for a better understanding of the new estimates.
ALERT™ subscribers can also download event sets and loss-based similar stochastic event (SSE) IDs for Hurricane Maria and a wind shapefile of the median loss event.
Please note that the new event set comprises five events. Four of these (Event 2, Event 3, Event 4, and Event 5) represent our revised estimated range of losses. The fifth (Event 1) represents storm parameters as reported by the NHC in real time; it is AIR’s opinion that there is sufficient evidence that the surface-level winds from Hurricane Maria are better reflected with the parameters of Events 2 to 5. We are providing Event 1, however, for purposes of sensitivity-testing and to emphasize that the losses could have been even higher. We urge clients to run the Touchstone® event sets (Events 2 to 5) provided on the ALERT website against their own portfolios, as that will provide a better evaluation of the potential impacts to their books of business.
Re-Analysis of Winds from Hurricane Maria Over Puerto Rico
A re-examination of the storm parameters used for the Hurricane Maria ALERT event set released on September 25, augmented by information that has only recently become available, called for modifications to the peripheral pressure (Pe), which represents the outer edge of the circulation associated with a tropical cyclone; central pressure (Cp) at landfall; and finally, the radius to maximum winds (Rmax). The modifications, which are described below, have led to a reduction in modeled wind speeds over Puerto Rico. For a more detailed discussion of our findings, please download Hurricane Maria – Understanding AIR’s Revised Modeled Loss Estimates from the Downloads tab.
Peripheral Pressure
In most parametric wind speed models, including AIR’s, the difference between the peripheral pressure, Pe, and the central pressure, Cp, is used as a proxy for the maximum wind speed (Vmax) at the radius of maximum winds (Rmax). The difference between Pe and Cp not only affects Vmax, but also the winds at all other locations within the storm’s wind field. In AIR’s Caribbean hurricane model, Pe is a constant value of 1013 mb, which well represents average atmospheric conditions in this region of the Atlantic during hurricane season. For any given storm however, this value can differ by several millibars (mb) in either direction. Inspection of surface pressure conditions at the time of Maria’s passage through the Caribbean revealed the Pe to be 1008 mb. The lower Pe effectively reduces the winds because it reduces the pressure difference between the center and the periphery of the storm. For an intense storm such as Maria, a reduction of 5 mb for Pe yields approximately a 5 mph difference in Vmax.
Central Pressure
We obtained a recently published report (Morgerman, Josh. “iCyclone Chase Report: Hurricane MARIA.” iCyclone. 20 September 2017. http://www.icyclone.com/upload/chases/maria/iCyclone_Chase_Report_MARIA2017.pdf.) that suggested the landfall Cp could also have been several mb higher than that reported by the National Hurricane Center (NHC) in real time. The landfall Cp for Maria from the NHC is an estimated value—based on reconnaissance aircraft—several hours before Maria made landfall. The central pressure value from the iCyclone Report was determined from the pressure trace taken by storm chaser Josh Morgerman who had positioned himself with two barometers just five miles north of where Maria actually made landfall. No other National Oceanic and Atmospheric Administration (NOAA) meteorological instruments are known to have survived the storm. (The NHC has incorporated iCyclone data in the past and they may do so with Maria data as well.) The two iCyclone instruments that were used are highly accurate and precise; in fact they were in agreement to within 0.1 mb for the time period observed, which instills further confidence in the reliability of their measurements. Using this information, we calculated the central pressure at the time of landfall to be 930 mb – about 7 mb higher than what was originally issued by NHC.
The combined effect of the higher central pressure and the lower peripheral pressure yields maximum wind speeds some 12-15 mph lower than in our original posting. Importantly, the winds at San Juan, where considerable exposure exists, were around 115 mph. These satellite-based wind speed estimates are from the Regional Analysis and Mesoscale Modeling Branch (RAMMB) of NOAA.
Radius to Maximum Winds
One additional piece of information we incorporated into the revised ALERT event sets was from a National Weather Service (NWS) radar image taken while Maria was just coming onshore. Although the radar was destroyed about two hours later, we used that image to determine that the Rmax was 19 miles. This value is close to but slightly less than those from the reconnaissance vortex reports (23 miles) and from RAMMB (21 miles). Importantly, we used all three values to bound the uncertainty for Rmax as well as the Cp estimate from NHC to bound the central pressure value at landfall.
AIR’s Damage Survey Findings
In late October, AIR Engineers performed a damage survey in Puerto Rico to understand the nature and extent of damage caused by Hurricane Maria across various lines of business, including residential, commercial, industrial, and auto. This section summarizes some of the key findings from AIR’s damage survey across different market segments.
What is important to note is that observations from the damage survey corroborated our revised view of Maria’s wind speeds, particularly in and around San Juan.
For a more detailed discussion and photos, please download Hurricane Maria – Understanding AIR’s Revised Modeled Loss Estimates from the Downloads tab.
Residential Buildings
Puerto Rico’s residential single-family home and low-rise apartment building inventory is dominated by reinforced concrete construction, colloquially referred to as bunker-style buildings. The floor and roof slabs are typically cast-in-place concrete with walls that are either reinforced concrete or reinforced masonry. While structural damage from Hurricane Maria to these floor and roof slabs was not evident, damage such as loss of wall siding, as well as breach of the envelope due to damage to windows and doors, and associated consequences with intrusion of wind and wind-driven rain was commonly observed. Failure of opening protection systems, damage to garage doors, outdoor lighting fixtures, and rooftop equipment were also common.
Mixed construction types among single-family homes were also common in areas away from the metropolitan area of San Juan. These had bunker style construction in the first floor, while wood or masonry construction completed the walls and roof structure in the second floor. Failures commonly associated with wood and masonry construction involving improper or insufficient connections, as well as internal pressurization-related failures, were seen in such homes.
Commercial Buildings
What stood out when it came to the performance of commercial buildings in Puerto Rico was the extent of damage to mid- and high-rise condominiums in San Juan that faced the water. While structural failure was not seen, major damage was evident in other forms:
- Damage to roofs and roof covers, which led to the intrusion of wind-driven rain into the interior, leading to contents damage and associated business interruption
- Damage to roofs caused by the failure of rooftop equipment anchorage
- Damage to soffits
- Loss of wall siding (typically stucco) applied over cast-in-place concrete
- Damage to windows and sliding glass doors, both of which had the same consequences associated with damage to roofs mentioned above
- Failure of opening protection systems, including shutter panel loss, shutter panel displacement, shutter track loss, and blowout of window to which shutter was attached
The damage described above was not seen only in condominium buildings, but was also prevalent in commercial buildings that house offices, hospitals, banks, car dealerships, restaurants, retail stores, and shopping malls, to mention a few.
Industrial Buildings
AIR’s definition of industrial buildings applies to a wide variety of industries and structures. Manufacturing alone accounts for about 50% of the island’s gross domestic product (GDP); however, manufacturing comprises not only life sciences (bio-pharma), but also aerospace, food processing, and rum, to mention but a few. In addition to manufacturing, the industrial base includes information technology, export services, food distribution, and creative services, among others.
All the damage mechanisms described in the previous section for commercial buildings were seen also in industrial buildings. The fundamental differences were the shape and size of the industrial buildings themselves or the assets within the industrial campuses. Unlike commercial buildings, industrial buildings were more commonly low-rise with higher ceilings. The presence of metal on the walls and roofs was also common, while a significant number of these buildings were also of reinforced concrete. Pharma manufacturing facilities, while built to stronger codes, exhibited a range of damage, from minimal to major for those near the landfall location. However, AIR has received information from the market and from our clients that losses to the pharma segment will be contained.
While physical damage to industrial buildings was seen in multiple cases, what remains uncertain is the level of losses to the individual coverages: building, contents, and business interruption (BI). Given our lack of access to the interior of the facilities in many cases, there is uncertainty regarding the level of loss to building interiors, mechanical equipment, including heavy machinery, and contents. While many facilities were up and running in some capacity due to the widespread presence of on-site power generation, including backup generators, it is undoubted that BI-related claims and losses will continue to develop for months to come until the island’s power grid is completely restored. Currently, they are relying on backup systems, and these are not meant to be run full time.
New Learnings on Insurance
Based on market feedback, AIR undertook a review of insurance practices and policies in Puerto Rico. What follows is an overview of our learnings.
Residential Policies
Residential insurance policies in Puerto Rico generally take two forms: homeowners and dwelling. The former includes coverage for contents and additional living expenses, while the latter does not. Dwelling policies, which are mortgage-based, are far more prevalent than dwelling policies in Puerto Rico, although the precise split is unclear. Also uncertain is the extent to which properties that do not carry mortgages are insured. Finally, there remains uncertainty around whether coverage on the primary buildings, which are typically bunker style, extends to wooden roofs or wood frame second stories.
Business Interruption and Commercial and Industrial policies
On October 26—Day 36 of the blackout on Puerto Rico—Maria was credited with causing what was termed the largest blackout in U.S. history. As of December 4, 62 out of 78 municipalities had their power restored, according to the Puerto Rico Electric Power Authority (PREPA)—known locally as Autoridad de Energía Eléctrica—but this number was caveated by the authority, which stated that it was subject to change due to the fragility of the system. The extent and duration of the impact of Puerto Rico’s lack of power on business interruption losses are uncertain. For example, most industrial facilities remain operational, albeit at lower levels of production due to having to run on generators. In many cases, the purchase of generators was a prerequisite for obtaining business interruption coverage. In particular, many large pharmaceutical facilities have secondary power generation capabilities to provide resilience in the face of a frail power generation and distribution system under PREPA. In addition, many of these facilities may have had sufficient stock on hand, which would limit their losses; others may not have such stores. Although it is clear that business interruption will play a role in the insured loss total for Puerto Rico, how large a role remains highly uncertain.
In addition, we have received consistent market feedback that commercial and industrial deductibles are commonly higher than the average of 2% that we utilize in deriving industry loss estimates. Furthermore, we have learned that many large industrial facilities’ insurance policies include a captive insurance layer, which means that they are at least partially self-insured.
To account for our enhanced understanding of market practices and insurance coverage in developing our industry loss estimates for Hurricane Maria, we have made a one-time adjustment to our take-up rates for Puerto Rico, from 50% to 40% for residential lines and 90% to 80% for commercial and industrial lines.
Other Sources of Uncertainty
Note that AIR’s modeled insured loss estimates include “demand surge,” which is the increase in the cost of labor and materials that is often observed in the aftermath of major catastrophes. In estimating insured industry losses for Maria, we have used our default demand surge function, which contributes roughly 20% to the total loss at the high end of the range. However, there is still uncertainty about the extent to which demand surge will impact losses from this storm as well as reconstruction efforts, given the blackout (see “New Learnings on Insurance”).
Additional uncertainty derives from not knowing whether a workforce large enough to handle the sizable amount of work that needs to be done is available and able to go back to work. According to the Governor of Florida—on average—2,000 people have migrated from Puerto Rico to Florida alone every day since Maria struck. More have migrated elsewhere in the Mainland U.S. Whether some or all migrate permanently to the Mainland U.S. and what percentage of those who have migrated were members of Puerto Rico’s workforce are unknowns.
In particular, it is uncertain whether there are enough claims adjusters on Puerto Rico to support claims settlement; enough people to work on the reconstruction effort; and enough people to work for the many businesses related to the tourist industry, such as hotels, restaurants, resorts, and the businesses that cater to them. A workforce hollowed out by migration to the mainland could further complicate getting businesses up and running and extend business interruption (BI)—not only complicating claims for BI but also generally impacting how well Puerto Rico’s economy and tourism industry rebound from Maria.
Claims will continue to develop over the course of the next six to eight months, largely as a result of continuing BI losses. While hotels had hoped to be open for business by Christmas—given Governor of Puerto Rico Rosselló’s claim that 95% of the island would have power by mid-December—it is unclear whether this claim will become reality and for how many hotels a return to business is feasible.
AIR’s modeled insured loss estimates include:
- Insured physical damage to onshore property (residential, commercial, and industrial) and auto due to wind and precipitation-induced flooding
- Insured loss to contents for homeowners policies only (not dwelling policies, see “Residential Policies” section), and commercial and industrial policies
- 2017 indexed take-up rates
- Losses due to business interruption
- Losses to industrial facilities
- Additional living expenses (ALE) for residential claims
- For residential lines in the U.S. territories of Puerto Rico and the U.S. Virgin Islands, 10% of modeled precipitation-induced flooding damage under wind policies
- For residential lines in territories other than in the U.S. territories of Puerto Rico and the U.S. Virgin Islands, 100% of flood losses
- For commercial lines in the U.S. territories of Puerto Rico and the U.S. Virgin Islands, insured physical damage to structures and contents and business interruption directly caused by precipitation-induced flooding, assuming a 10% take-up rate for commercial flood policies
- For commercial lines in territories other than in the U.S. territories of Puerto Rico and the U.S. Virgin Islands, 100% of flood losses
- For business interruption losses, direct and indirect losses for insured risks that experience physical loss
- Demand surge (see “Sources of Uncertainty” section)
AIR’s modeled insured loss estimates do not include:
- Losses to infrastructure
- Losses from hazardous waste cleanup, vandalism, or civil commotion, whether directly or indirectly caused by the event
- Losses to offshore properties, pleasure boats, and marine craft
- Losses resulting from the compromise of existing defenses (e.g., levees)
- Losses to uninsured properties
- Other non-modeled losses, including loss adjustment expenses
We invite you to register for our webinar, “Looking Back at an Active 2017 Hurricane Season,” for more details.