Earthquake Wildfire

advertisement
Managing Catastrophe Risks in
The Wild West: Earthquakes
and Wildfires
Dan Loris, Global Head of Property,
Engineering Lines & Marine for Zurich
Insurance
Agenda
• Earthquakes
– Overview
•
•
•
Earthquakes 101
Past events and key faults
Facts and myths
– Coverage
– UW Considerations
– Advances: Predicting and Warning
• Wildfires
– Historical Experience
– Exposure
– Underwriting Considerations
Earthquake 101—Plate Tectonics
• Most earthquakes are caused by the interaction of the Earth’s plate boundaries.
• An example would be the San Andreas fault, which separates the North
American Plate from the Pacific Plate, forming part of the “Ring of Fire.”
• Volcanic activity is closely linked to plate tectonics.
• Four types of plate boundaries:
•
•
•
•
Divergent—Plates pull away.
Convergent—Plates come together.
Transform—Plates slide by each other.
Boundary Zones—Interaction unclear.
Cross section by Jose F. Vigil, produced by the USGS, Smithsonian Institution and U.S. Navy
Earthquake 101—Types of Faults
Source: USGS
Earthquake 101—Cause and Effects
•
Earthquakes are the resulting ground motion we feel when plates that are locked by
friction under stress suddenly let go.
•
Elastic Rebound Theory:
•
Faults are locked
(through frictional
forces) while strain
energy builds up and
deforms crustal rock.
•
At some point,
frictional forces are
overcome and strain is
released.
The 1906 SF EQ caused 2.5m of
slip as captured by this fence offset
(Gilbert, USGS).
EQ Magnitude and Intensity
Comparison of EQ Magnitude and Intensity Scale
Richter
Magnitude
Equivalent
Energy in TNT
Equivalent
Energy in
Hiroshima Size
Atomic Bomb
3-4
15 tons
1/100th
II-III
Truck or train vibration
4-5
480 tons
3/100th
IV-V
Sleeper awoken
5-6
15,000 tons
1
VI-VII
Difficult to stand, masonry damage
6-7
475.000 tons
37
VII – VIII
General panic, some buildings
collapse
7-8
15,000,000
tons
1160
IX-XI
Wholesale destruction
8-9
475,000,000
tons
36,700
XI-XII
Total destruction, visible ground
amplification
M5
Mercalli
Intensity
M6
Impact
M7
For every increase in magnitude, there is an approximately 32 times increase
in energy released.
Top 10 Most Costly U.S. EQ’s Inflation Adjusted
Rank
Date
Location/Magnitude
Loss
When
Occurred
Insured
Loss When
Occurred
Insured Loss
Inflation Adj
(2013 dollars)
1
Jan 17, 1994
Northridge, CA M6.7
$44bn
$15.3bn
$24.05bn
2
Apr 18, 1906
San Francisco, CA M7.8
$524m
$180m
$4.24bn
3
Oct 17, 1989
Loma Prieta, CA M6.9
$10bn
$960m
$1,8bn
4
Feb 28, 2001
Olympia, WA
$2bn
$300m
$395m
5
Mar 27, 1964
Anchorage AK M9.0
$540m
$45m
$340m
6
Feb 9, 1971
San Fernando, LA M6.5
$553m
$35m
$200m
7
Oct 1, 1987
Whittier, CA
$360m
$75m
$155m
8
Apr 4, 2010
San Diego, LA
$150m
$100m
$105m
9
Sep 3, 2000
Napa, CA
$80m
$50m
$68m
10
Jun 28, 1992
San Bernardino, CA
$100m
$40m
$66m
Source: Munich Re, Geo Risks Research, NatCatSERVICE.
Top 10 Most Costly U.S. EQ’s Based on Current Exposure
Rank
Date
Location
Mag
Insured Loss
(Current Exposure)
1
Feb 7, 1812
New Madrid, MO*
7.7
$112 bn
2
Apr 18, 1906
San Francisco, CA
7.8
$93 bn
3
Aug 31, 1886
Charleston, SC
7.3
$44 bn
4
Jun 1, 1838
San Francisco, CA
7.4
$30 bn
5
Jan 17, 1994
Northridge, CA
6.7
$23 bn
6
Oct 21, 1868
Hayward, CA
7.0
$23 bn
7
Jan 9, 1857
Fort Tejon, CA
7.9
$ bn
8
Oct 17, 1989
Loma Prieta, CA
6.3
$7 bn
9
Mar 10, 1933
Long Beach, CA
6.4
$5 bn
10
Jul 1, 1911
Calaveras, CA
6.4
$4 bn
Source: AIR Worldwide, Insured losses in billions
For New Madrid (1811 to 1812) only one event selected.
San Andreas Fault (Facts & Myths)
•
•
•
•
•
•
•
The San Andreas is a transform boundary fault (strike – slip).
The Pacific Plate is moving north and the North American Plate is moving south.
Somewhat unique in that most plate boundaries occur below the ocean.
Movement can occur slowly and steadily (creep) or suddenly.
Locked segments can cause tremendous EQ’s due to build-up of stress.
– 1906 SF quake ruptured along a locked 430k km long segment.
Little or no concern for tsunamis from this fault.
Could CA “break off” and fall into the Pacific???
San Andreas Fault near San Luis Obispo
Photo by Robert E Wallace, USGS
Source: USGS Simplified Paper 1515
San Andreas Fault SF—Stress Release and Transfer
•
•
Total of 14 M6 or > events prior to 1906, only 1 post 1906 EQ
Time dependence (point in time relevant to events) important as frequency is
not a constant
Credit: Ross Stein, R. E. Crippen, USGS
San Francisco 1906 Earthquake (Magnitude 7.8)
•
•
•
•
•
•
•
Occurred on the San Andreas fault (transform fault), rupturing a 300 mile section.
Over 3,000 people died and 80% of the city was destroyed.
It is estimated that as much of 90% of the damage was attributable to the resulting fire.
Some property owners deliberately set fire to EQ damaged homes to collect insurance.
Of the 137 insurance companies involved, twenty went bankrupt.
Creation of the SFP came as a result.
Largest ever U.S. relief effort to this day.
Source USGS
SF City Hall, from Steinbruge Collection, UC Berkley EQ Engr Research Center
Northridge EQ (Magnitude 6.7)
•
•
•
•
•
Occurred Jan 17 1994; currently 5th largest insured Cat loss in U.S. History ($24 bn)
Was caused by the Santa Monica Mountains Thrust Fault 20 miles NW of downtown LA
Most damage to Santa Monica, Simi Valley and Santa Clarita
At M6.7 was a moderate event lasting only 10 to 20 seconds but had high ground
acceleration, felt as far away as Las Vegas
Occurred on a blind fault, meaning the fault does not break the ground surface making
identification very difficult
Cascadia Fault
•
A convergent plate boundary “mega thrust” fault stretching from Vancouver to Northern CA
•
The Juan de Faca plate subducting beneath the NA plate, forming a long fault line
•
Last major event 1700 AD, with an average RP of 300 to 600 years (long tail)
•
Can produce M- 9.0 or greater events with extreme tsunamis
•
Many volcanoes along the fault
•
10% to 15% chance of a M-9 or higher event in next 50 years
•
30% chance of a M-8 or higher in same period
•
Some study into relationship with San Andreas fault
New Madrid EQ Risks—Ground Attenuation Implications
•
•
•
•
•
•
•
A total of 4 magnitude 7.0 to 8.1 earthquakes make up the 1811 to 1812 events.
Felt over an area of 50,000 sq. mi., compared to only 6,200 for the 1906 SF EQ.
This is due to the differences in ground attenuation.
Underlying cause of the quake not well understood but believed to be the result of a rift
(scar) when the NA plate began to breakup about 750 million years ago.
Forecasters estimate a 7% to 10% chance of a 7.5 M to 8.0 M event in the next 50 yrs.
At the time, the areas hardest hit were not inhabited, today Memphis is severely
exposed.
New USGS map, released July 2014, expands the critical zone.
Source: USGS, New Release Map as of July 2014
Source: Wikimedia Commons
Source: SLU Earthquake Center
Who Provides EQ Insurance in the U.S.?
Top Ten Largest Writers of EQ in the U.S.
Rank
Group/Company
Direct Premiums
Market Share
1
CA EQ Authority
$566.7m
19.9%
2
State Farm
$231.9m
8.1%
3
Zurich (excludes Farmers)
$193.2m
6.8%
4
AIG
$183.6m
6.5%
5
Travelers
$137.8m
4.8%
6
GeoVera
$121.0m
4.3%
7
Axis
$102.8m
3.6%
8
Liberty Mutual
$94.0m
3.3%
9
Ace
$78.9m
2.8%
10
Swiss Re
$65.7m
2.3%
Data at year end 2012. Source SNL Financial LC
Who is the California EQ Authority?
•
Formed in 1996 out of need specifically for CA after the 1994 Northridge EQ
•
Insures homeowners, condo owners and renters (no commercial), rated A- (excellent)
•
Sold through participating insurance companies
•
Publically managed and privately funded, not-for-profit
•
Currently collects approximately $570m annually in DWP
•
Has claims pay capacity of $10.2 bn
•
35% market share in CA, 800k policy holders, avg. $750/policy
•
Deductibles from 10% to 15% of dwelling limit
U.S. EQ—Risk Versus Reward (a Capital Driven Exposure)
U.S. Direct Written Premium by Year for EQ (in billions)
3
$2.58
2
$1.82
$1.34
1
$1.43
$1.05
$1.21
$1.30
$1.38
98
99
0
$1.93
$2.06
$2.19
$2.49
$2.57
$2.66
$2.13
$2.04
$2.26
$1.55
$0.94
0
94
95
96
97
1
2
3
4
5
6
7
8
9
10
11
12
•
Source A.M. Best & Insurance Information Institute
•
Significant capital required to write EQ (compare 1994 Northridge insured loss of
$24.1bn to average countrywide premiums of approximately $2bn annually)
EQ premium of $2.26bn is well off its 2010 peak
CA (including the CEA) represent almost ½ of countrywide premiums
Take up rate in CA now 12%, down from 30% in 1996
By region 22% of HO’s in West buy, 11% in MW, 6% in S and 5% in NE
In last ten years, combined ratios have varied from as low as 30% in 2007 to 55.8% in
2011
•
•
•
•
•
EQ Coverage
•
•
•
•
•
•
Earthquake is a required offering and typically provided through the CEA for
homeowners insurance in CA.
For commercial (particularly large), EQ is typically added by endorsement and is nearly
always subject to a sublimit and annual aggregate.
Deductibles are typically expressed as % of value for PD and % or time period for Time
Element.
Loss resulting from fire (earthquake fire following) is typically covered at policy limits and
subject to the fire deductible even if clearly determined to be a result of an EQ. CA is a
statutory fire policy state (SFP).
DIC (difference in conditions) coverage is usually peril specific and is intended to cover
only direct shake damage and often excludes secondary causes of loss such as EQ FF.
Litigation in CA has proven to be challenging for insurance carriers.
–
•
•
For example, Senate bill 1899, Code of Civil Procedures 304.9 re-opened a large number of
closed 1994 Northridge EQ claims in 2001 (seven years after the event).
Commercial policyholders are often afforded earth movement coverage, which in
addition to EQ, may include volcanic activity, subsidence, landslides etc.
In the U.S., loss from tsunami is often covered as flood even if resulting from an EQ.
UW Considerations
Many considerations go into the underwriting of earthquakes. These may include:
•
•
•
•
•
•
•
•
•
•
•
Location (seismic zone, proximity to known faults)
Construction type
Soil conditions (soft, unstable and fill tend to amplify shaking and are subject to liquefaction)
Human element (applicable for commercial)
Occupancy
Ensuring loss from secondary perils
Overall portfolio accumulation
Cost of capital and reinsurance
Adequacy of premium
Legal and regulatory environment
Confidence in model (relative to frequency models like U.S. HU)
–
–
•
•
•
Time dependence (near-term vs. long-term return period average)
Capability to account for loss amplification
Ability to handle influx of claims
Longevity of claim payouts “tail” relative to hurricane
Historical experience is typically not a key factor in most cases due to EQ’s low frequency/high severity
UW Considerations: Secondary Causes of Loss
•
Fire following: ruptured gas lines compounded by impairments to fire protection
•
Water damage: due to broken sprinkler and domestic piping
•
Tsunamis: particular concern for PNW (OR, WA, AK, YVR)
•
Flood: due to collapsing or breaking of dams or dikes
•
Civil commotion: including riot, theft and vandalism
•
Liquefaction: inability to rebuild, landslides (ground destroyed)
Portfolio Management
•
•
•
Earthquake is a capital intensive peril due to its low frequency but high severity, so
strong portfolio management is of particular importance.
Goal is to spread risk as evenly as possible across the peril region.
Challenge with CA is that exposure is concentrated in three major metro areas:
–
–
–
Los Angeles (50%)
San Francisco (30%)
San Diego (20%)
•
Compare this to U.S. wind, where
exposure is diversified across the entire
Gulf and Eastern seaboard.
•
Not just a property exposure, need to
consider workers comp for example.
•
Opportunities (where rating plans allow)
include varying price to incent take-up in
lower accumulation areas and varying
limits offered to impact PML’s.
EQ Reinsurance Insights
•
•
Many U.S. Cat treaties include CA EQ in addition to U.S. wind and other perils
RE capacity for EQ is currently readily available
–
•
•
However, adequacy is closely tied to U.S. wind.
After 2005 Hurricane Katrina, EQ reinsurance pricing spiked.
Many E&S writers impacted by Katrina also faced limitations on ability to write CA
EQ.
Man Made Earthquakes—Fracking
What is Fracking?
•
•
•
•
It is a process by which drillers can obtain natural gas deep below the surface.
Sand, water and chemicals are pumped into wells to induce movement in deep rock formations.
The high pressure fractures the rock and allows the trapped gas to flow into the wells.
Waste water used in the process is then normally injected underground.
How does Fracking induce EQ’s?
Underground pore pressure is increased when waste water is injected. This can lubricate nearby
faults and can cause slippage, which releases stress causing an earthquake.
What is the consensus?
•
•
•
The potential impact is widely debated, such
as the case involving a 2011 M5.6 EQ in
Prague, OK, and a 5.3 quake in Colorado
near a fracking operation.
Fracking often causes magnitude 2 to 3 EQ’s.
The USGS remains somewhat agnostic on
the issue, but several states in the U.S.
intend to begin mapping and recording
events thought to be related to fracking.
Predicting Earthquake
•
The fact is no one can predict when or where a earthquake will occur with enough
certainty to allow for immediate action to be taken prior to an event.
•
Time dependence and stress transfer are a focus of work being performed at the USGS.
Data courtesy of Ross Stein USGS
Data courtesy of Ross Stein USGS
Earthquake Fault Mapping & Early Warning
•
•
•
Many faults, such as the one that caused the 1994 Northridge EQ, were widely unidentified
prior to the events.
On July 2, 2014, CA Governor Jerry Brown approved a plan to identify hidden faults and to
create zoning maps that would identify them and potentially even restrict future development.
Earthquake early warning involves detecting seismic waves.
–
–
•
•
•
•
P waves are the fastest and move at 15,000 mph but do little damage.
The more damaging S waves move more slowly.
Sensors like the ones employed by the U. of Berkley would detect S waves then sound an
alarm giving up to one to two minutes time to prepare before the P waves arrive.
Currently, there are more than 400 sensors employed in CA; systems also exist in Japan.
Funding in the U.S. remains a challenge.
Opportunities exist to speed communication.
Wildfire
Wildfire—Experience Tells Us:
•
Annual average of 140,000 brushfires took place in the U.S. from 1916 to 1996.
•
15,000 homes and other insured properties in the U.S. were destroyed by brushfires
from 1985 to 2003.
•
Structures are either completely destroyed or only suffer slight to moderate damage;
there is very little in between.
•
There has been an increase in insured losses from brushfires, which has been primarily
driven by the increase in the number of exposed properties in high risk areas.
•
•
•
More than 7.2 million homes are located within the three highest risk areas in California, which
include Los Angeles County, Alameda County, and San Diego County.
Because the majority of brushfires have affected
CA, the primary focus on modeling has been CA
for most firms.
To date, brush or wildfire have impacted more
personal lines and small business carriers than
large commercial.
The Ten Most Costly CA Wildfires (in billions)
Rank
Date
Location
Loss When
Occurred
In 2012
Dollars
1
Oct. 20–21, 1991
Oakland Fire
$1.70
$2.70
2
Oct. 21–24, 2007
Witch Fire
$1.30
$1.35
3
Oct. 25–Nov. 4, 2003
Cedar Fire
$1.10
$1.24
4
Oct. 25–Nov. 3, 2003
Old Fire
$.975
$1.14
5
Nov. 2–3, 1993
LA County Fire
$.375
.60
6
Oct. 27–28, 1993
Orange County Fire
$.350
.52
7
June 27–July 2, 1990
Santa Barbara Fire
$.265
.48
8
Sept 22–30, 1970
Oakland Hills Fire
$.025
.138
9
Nov. 24–30, 1980
LA, Orange County
$.043
.112
10
July 26–27, 1977
Santa Barbara,
Montecito
$.020
.071
Source: ISO’s Property Claim Services Unit (PCS)
Photo by Mark Thiessen
Historical Frequency (by Month) for CA Brushfire
California Brushfire Frequency by Month
30%
2003 & 2007 wildfire sieges
occurred later in season and
were larger due to Santa
Ana Conditions and dry
vegetation.
25%
20%
15%
10%
5%
0%
Jan
Feb
Mar
Apr
Source Aon Benfield Impact Forecasting
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Historical Brush Fires in CA
Source: Aon Benfield Impact Forecasting (With Permission)
Surface Fuels in CA
Source: Aon Benfield Impact Forecasting
Winds: Santa Ana, Sun Downer & El Diablo Foëhn Winds
Föehn Winds:
Seasonal, strong
downward sloping
winds in the
mountains. A
combination of
prevailing highpressure system
in the Great Basin
and a lowpressure system
off the coast of
California.
Underwriting Brushfire Risks
•
Underwriting exposures
– Accurate location data (terrain including slope)
– Construction of house and roofing
– Brush clearance
– Size of eaves
– Grade of roof shingles
– Double pane vs. single pane windows
– Extinguishing systems
– Fire department capabilities
– Contingency plans (commercial)
– Urban conflagration potential
•
Data capture
– More detailed data available, the better model performance
•
Modeled losses vs. recovery from Cat cover
In Summary:
Earthquake:
• EQ magnitude is just a measure of how much energy is released, but many variables
factor into how severe an event may be.
• A lot is known about what causes earthquakes, however much is yet to be discovered.
• No one can accurately predict exactly when or where an earthquake may occur.
• EQ cover is challenging to underwrite due to its high severity, low frequency nature.
• Good UW requires strong knowledge of portfolio accumulation and good data.
• Secondary perils, such as fire following, create significant challenges.
• Excluding EQ doesn’t mean an insurer isn’t still significantly exposed to an EQ event.
• EQ early warning systems provide some hope of alerting the population of pending loss.
Wildfire:
• Over the last 20 years, there has been an increase in the occurrence of wildfires in CA.
• Dry conditions and encroachment into rural areas has contributed to loss amounts.
• Historical losses have been driven more by residential than commercial exposures.
• Underwriting the risk is straightforward but information intensive.
Managing Catastrophe Risks
in the Wild West:
Earthquakes and Wildfires
Tim Smail
Senior Vice President of Engineering and
Technical Programs
Agenda
• Federal Alliance for Safe Homes (FLASH)
– Who are we?
• Earthquakes
– History of earthquake building codes
– Beyond code programs/examples
– Tools for communities
• Wildfires
– History of wildfire building codes
– Beyond code programs/examples
– Tools for communities
Mission: Strengthening Homes and Safeguarding
Families From Disaster of All Kinds
• At FLASH, we partner with leading public, private and
nonprofit academic, consumer, entertainment, financial
services, product, research, service and technical
organizations to deliver the latest advances in disastersafety information to the public.
• Create a public value for strong, safe and sustainable
homes.
• Deliver initiatives that fit into two program tracks:
– Storytelling for the public
– Curriculum for students and professionals
• Mainstream the science of safety.
Partners in Prevention: 16 Years of Growth
The Disaster Safety Movement: 16 Years of Impact
Convening a Conversation for Safe, Sustainable
and Resilient Communities
Partnership With Leading Organizations for Disaster
Safety and Resilience
Brief History of Earthquakes
• From 1769 through 1994, major earthquakes registering 6.3 to 8.3
on the Richter scale were experienced every 5.4 years on average
in California.
Loma Prieta Earthquake, 1989
Santa Rosa Earthquake, 1906
History of Seismic Building Codes
•
•
Santa Barbara (local involvement)
–
Earthquake on June 29, 1925.
–
February 1926, City Council agreed to require structures to be designed to resist
horizontal forces.
–
Palo Alto City Council adopted an amendment to its building code (design to
withstand lateral acceleration) eight months later.
California (state involvement)
–
1933 Long Beach earthquake.
–
115 deaths, $60 million in damage.
–
70 schools destroyed, 75% of schools were heavily damaged.
–
Soon after, California enacted the Field Act, seismic design forces for school
buildings, and Riley Act, seismic design for most public buildings in California.
Source: State of California Seismic Safety Commission. 2000. A History of the California Seismic Safety Commission,
www.seismic.ca.gov/pub/CSSC_HISTORY.pdf; FEMA. 2010. FEMA P-749. Earthquake-Resistant Design Concepts: An
Introduction to the NEHRP Recommended Seismic Provisions for New Buildings and Other Structures, https://s3-usgov-west-1.amazonaws.com/dam-production/uploads/20130726-1759-25045-5477/fema_p_749.pdf.
Seismic Safety: A Modern Approach
• Alaskan earthquake of 1964
– Stanley Scott, seismic-safety policy analyst with the Institute for
Governmental Studies at University of California (UC), Berkeley.
– Attended a conference to review the Alaskan earthquake and how it
related to public safety in California.
– Karl Steinbrugge, School of Architecture at UC, Berkeley, published
a paper, at Scott’s request, suggesting that there are mitigation
options for earthquakes.
– August 8, 1969, a Joint Committee on Seismic Safety was created
with a goal to “develop seismic safety plans and policies and
recommend to the Legislature any needed legislation to minimize the
catastrophic effects upon the people, property and the operation of
our economy should a major earthquake strike any portion of the
state of California.”
Modern Seismic Safety: 1971 San Fernando Earthquake
• During 1971 and 1972, 35 pieces of legislation were offered.
– Seismic Safety General Plan Element: required city and county
plans to include a seismic-safety event
– Strong Motion Instrumentation Program: required earthquake data
gathering equipment to be installed in large buildings and be
monitored
– Dam Safety Act: required OES and public safety officials to order
dam owners to prepare emergency evacuation plans and maps of
areas that would be flooded
• In 1972, Governor Ronald Reagan established the Governor’s
Earthquake Council.
• The council existed until the Seismic Safety Commission was
established in 1975.
Seismic Safety: A National Approach
•
1927, first seismic regulations as voluntary appendix in 1927 Uniform
Building Code.
•
Congress passed Earthquake Hazards Reduction Act of 1977. Established
National Earthquakes Hazard Reduction Program (NEHRP), with
responsibilities for FEMA, EMA, NIST, NSF and USGS.
–
–
These groups are focused on mitigating earthquake risks to the national
economy and life/safety of building occupants.
Examples of NEHRP functions:
•
•
–
USGS—identifying level of earthquake hazard throughout the U.S.
NEHRP Recommended Provisions and Design Guidance  National Model
Building Codes, Standards and Guidelines
Since the 1997 edition of the NEHRP Recommended Provisions, both the
IBC and ASCE/SEI 7 Standard have based their seismic design criteria on
the latest recommendations in the Provisions recommendations.
Source: FEMA. 2010. FEMA P-749. Earthquake-Resistant Design Concepts: An Introduction to the NEHRP
Recommended Seismic Provisions for New Buildings and Other Structures. Available: https://s3-us-gov-west1.amazonaws.com/dam-production/uploads/20130726-1759-25045-5477/fema_p_749.pdf.
Seismic Safety: A National Approach
•
NEHRP Recommended Provisions describe the goal of the design
standards as follows:
–
“If subjected to sufficiently strong ground shaking, any structure will collapse.
The goal of the Provisions is to provide assurance that the risk of structural
collapse is acceptably small, while considering that there are costs
associated with the designing and constructing structures to be collapseresistant. The Provisions defines a reference earthquake shaking level…and
seeks to provide a small probability that structures with ordinary occupancies
will collapse when subjected to such shaking. The acceptable collapse for
structures that house large numbers of persons or that fulfill important
societal functions is set lower than this and additional objectives associated
with maintaining post-earthquake occupancy and functionality are added.”
Source: FEMA. 2010. FEMA P-749. Earthquake-Resistant Design Concepts: An Introduction to the NEHRP
Recommended Seismic Provisions for New Buildings and Other Structures. Available: https://s3-us-gov-west1.amazonaws.com/dam-production/uploads/20130726-1759-25045-5477/fema_p_749.pdf.
Seismic Safety: A National Approach
•
•
•
Seismic provisions in the International Building Code (IBC) and International
Residential Code (IRC) published by the International Code Council through
a consensus process.
Seismic codes are intended to protect people inside buildings by preventing
collapse and allowing safe evacuation.
Perhaps the best evidence of the efficacy of seismic codes are their
performance in past earthquakes:
–
Loma Prieta earthquake—1989
•
–
Few collapsed buildings for those built to seismic code, and most damage limited to
unreinforced masonry built before seismic codes adopted.
Northridge earthquake—1994
•
Almost all buildings built to seismic code remained standing and permitted
occupants to safely escape (although there were several tragic exceptions).
Source: FEMA. 1998. FEMA 313, Promoting the Adoption and Enforcement of Seismic Building Codes: A Guidebook
for State Earthquake and Mitigation Managers.
Seismic Safety: A National Approach
•
While better seismic construction practices were adopted in the western
U.S., as discussed, in the mid-1970s, these codes did not come into use in
the eastern U.S. until the 1990s.
•
Adherence to seismic codes is not as expensive as many people believe.
– 1% to 2% of total cost on new builds
– Higher on retrofits
•
It is the responsibility of local governments to adopt, enforce and keep
current building codes.
Beyond Code Examples: FLASH
Beyond Code Examples: Brace + Bolt
Beyond Code Examples: IBHS and DHS
• Fortified for Safer Living
– The FORTIFIED program specifies construction, design and
landscaping guidelines to increase a new home’s resistance to
natural, catastrophe-level perils most likely to occur in the area
where the structure is located.
• Resilience STAR
– The U.S. Department of Homeland Security (DHS) is creating a new
program called Resilience STAR™ to build and retrofit homes that
are more disaster resistant. The first phase of the pilot project will
focus on single-family homes in hurricane-prone communities.
Earthquake Tools for Communities
History of Wildfire
• 1985—The Wake-Up Call
San Diego, California
Horry County, South Carolina
History of Wildfire Building Codes: NFPA
• Firewise: National Fire Protection Association
– 1985-1986
•
•
1,400 homes burned down due to wildfires
Federal agencies tasked with wildfire response partnered with NFPA to
address the issues facing the wildland-urban interface
– 1987-1993
•
Started focus on reaching firefighters and homeowners with safety
information and, later, specific advice for homeowners and landscapers
– 1994-1998
•
Development of the Firewise website and using research
– 1999-2002
•
•
Firewise community planning workshops started
Pilot program called “Firewise Communities/USA” initiated
– 2003 to present
•
Firewise Communities® Recognition Program (>1000 communities)
Source: Firewise at NFPA: A brief history. Available: http://www.firewise.org/about/history.aspx.
History of Wildfire Building Codes: ICC
• Urban-Wildland Interface (WUI) Code
– 2003 initial issue
•
Supported with FEMA Grant Program, administered by the California
Governor’s Office of Emergency Services
– Initiatives address planning, designing, building and retrofitting
homes, as well as properly maintaining them and their property
– The following mitigation measures are addressed:
•
•
•
•
•
•
•
Roof
Walls
Soffits
Openings
Decks
Water supply
Defensible space
– Important to remember, local jurisdictions must adopt and enforce
the WUI Code for it to work!
Beyond Code Examples: FLASH
• Blueprint for Safety
Roof
Openings
Assembly
Class
Additional recommendations
Roof covering assembly should be
based on the roof slope and
fire rated. *Note: Class A rating is
type of covering can be found in
best for homes in extremely high
Home Builder's Guide to
prone wildfire regions. Wood
Construction in FEMA Wildfire
shake
shingles
are
not
Zones - Technical Fact Sheet
recommended even if fire rated.
No. 5.
FEMA, BFS, FSL, WUI
Class "A", "B", or "C" - UL
790 - Tests for Fires
Resistant Roof Covering
Materials or ASTM E-108 Standard Test Method for
Fire Tests of Roof Covering
Windows,
Doors,
Skylights
Exterior windows must be doublepaned glass and include resistant
screens or non combustible
shutters are required. Exterior
doors with windows and skylights
must also be double paned,
tempered glass.
FEMA, BFS, FSL, WUI
N/A
Minimize the size and number
of windows on the downhill side
of the house or side most likely
to be exposed to wildfire. Solid
metal shutters can provide
protection for windows and
sliding glass doors in a wildfire.
Wildfire Tools for Communities
Managing Catastrophe Risks
in the Wild West:
Earthquakes and Wildfires
Glen Daraskevich, SVP
Karen Clark & Company
Agenda
• How catastrophe models are built and estimate loss
– Components of an earthquake model
– Estimating potential losses
– Differences in wildfire and earthquake modeling
• Challenges with models
• Advancements in risk modeling?
• Deriving benefits from risk modeling
Catastrophe Models Simulate Thousands
of Hypothetical Events to Estimate Loss
Exposure-Data Input
Inventory of buildings and people including location
(street address) and physical characteristics (building
materials, size, value, etc.).
Catastrophe Model
Event Generation
Hazard
Intensity
Create a large sample
of hypothetical
events
Where? How big?
How frequent?
For each event,
calculate intensity at
each location
Vulnerability
Financial
Apply policy
conditions
to estimate
insured losses
Based on intensity
and exposure at
each location,
calculate damage
Sim Year
Event ID
Loss ($ million)
Loss Estimate
Output
1
1
253
Statistics and estimates of
hypothetical loss scenarios,
including expected loss and
large loss scenarios.
1
2
41
2
3
5
3
4
27
.
.
.
How Loss Estimates Are Used to Calculate
Traditional Risk Metrics
Ranking the loss estimates in
descending order creates the
exceedance probability curve
Rank
Return
Period
(Years)
Exceedance
Probability (EP) Curve
Probability
p(L) that
losses will
exceed L
Sim Year Loss ($ million)
1
10,000
838
2,478
2
5,000
76
2,068
3
3,333
1407
1,836
4
2,500
6234
1,623
…
…
…
100
100
587
…
…
…
1.0%
•
1 in 100
0.4%
1 in 250
•
Loss, L
…
341
Probable Maximum Loss (PML)—Point on the EP Curve
…
Sum of Loss
10,000
Average Annual Loss (AAL)—Expected loss from EP Curve
What Do Scientists Know About the 18111812 New Madrid Earthquakes?
• A violent shock of an earthquake
was accompanied by a very awful
noise resembling loud but distant
thunder.
• Complete saturation of the
atmosphere with sulphurious
vapor causing total darkness… .
• The cries of fowls and beasts of
every species and the crackling of
trees falling… .
• The roaring of the Mississippi… .
From Eliza Bryan’s personal account in Lorenzo Dow’s Journal,
published by Joshua Martin in 1849.
Whatever We Know About the Damage Is
From Newspaper Accounts
There Is Scientific Disagreement on the Magnitudes of
the New Madrid Earthquakes and the Return Periods
USGS
1996 Report*
USGS
2002 Report*
1,000 years
500 years
Recurrence Interval
Recurrence Interval
M 8.0 (1.0)
M 7.3 (0.15)
M 7.5 (0.20)
M 7.7 (0.50)
M 8.0 (0.15)
* Magnitudes (Weights)
Under Strong Ground Shaking, Saturated, Sandy Soils
Experience Liquefaction and May Create Sand Blows
Source: USGS
Strong shaking of the saturated sandy materials restructures the sand particles and causes an increase in the pore water pressure.
As the pore pressure becomes greater than the weight of the overburden materials, the soil liquefies and, in some cases, creates
sand blows.
Paleoliquefaction Studies Help Estimate Return Periods
From Tuttle and Schweig: Prehistoric Liquefaction Features
Where Will the Fault Rupture?
Three possible fault scenarios (five in USGS 2008)
Microseismicity helps identify
shape of fault
Source: USGS
Ground Motion Is Estimated Using Attenuation Equations
log(Y) = c1 + c2M + c3log(R + c4) + c5R + S + σ
Where:
Y=Ground motion or spectral values
M=Magnitude
R=Source–site distance
S=Site factor
ci=Coefficients obtained empirically
from observed or simulated GM data
σ=GM uncertainties
Simulating a Footprint and Impacts From Local Soil Conditions
Ground Motion From Attenuation Equations
Ground Motion Incl. Local Soil Conditions
Estimating Damage
• Vulnerability functions are mathematical relationships between ground
motion and expected damage.
• Functions are developed based on claims data, structural analysis and
engineering opinion.
• Data are predominantly available for residential structures; limited
information on commercial buildings and high-rises.
Modeling Wildfires
• About 80,000 wildfires occur in the U.S.
annually, but about 95% are suppressed quickly,
and a small portion pose risks to structures.
• Homes in WUI at greatest risk.
• Modeling wildfires:
–
–
–
–
–
–
Frequency and location of ignition
Vegetation (fuels)
Topography (slope)
Weather conditions (winds)
Site conditions near property
Human intervention
• Kiss-or-kill peril
Sources: National Interagency Fire Center, Calfire
Differences in Wildfire and Earthquake Catastrophe Models
Earthquake
Wildfire
Frequency
Low
High
Event severity
High
Low
Intensity factors
• Distance to faults
• Soils/liquefaction
• Proximity to WUI
• Vegetation
• Seasonal weather
• Winds during event
Structure vulnerability
Construction materials
• Construction materials
• Local surroundings
• Fire-suppression efforts
Modeling challenges
Sparse data on past events:
• Magnitude/frequency
• Ground-motion footprint
Events heavily dependent on human
factors:
• Ignitions
• Fire suppression
Vegetation and WUI constantly changing
Fine resolution model
UW challenges
Building codes focused on structure
survivability, not limiting loss
• Local conditions heavily influence and fire
suppression key factors in a loss
•Local conditions change
Scarcity of Data Creates Significant Uncertainty in
Estimating Magnitude and Return Period for Events
USGS
1996 Report*
USGS
2002 Report*
1,000 years
500 years
Recurrence interval
Recurrence interval
M 8.0 (1.0)
M 7.3 (0.15)
M 7.5 (0.20)
M 7.7 (0.50)
M 8.0 (0.15)
* Magnitudes (weights)
Logic Tree for New Madrid Seismic Zone
(NMSZ) From the USGS 2008 Report
USGS: Documentation for the 2008 update of the United States National Seismic Hazard Maps
New Madrid Magnitude Estimates and
Logic Tree From 2014 USGS Report
Uncertainty in Ground-Motion Attenuation
Equations Significantly Affects Loss Estimates
Sampling Error Can Lead to Different Views of Risk
Random Events at 500-Year Return Period
Sample A
Sample B
Earthquake Modeling: New 2014 USGS Report
• Latest science from 2014 USGS National
Seismic Hazard Maps
–
–
–
–
–
2008—2% in 50 Years PGA
More multisegment ruptures in California
New ground-motion equations
Updated earthquake catalog
Updated fault catalog
New slip rates
Ratio 2014 Divided by 2008—2% in 50 Years PGA
2014—2% in 50 Years PGA
View of Earthquake Risk Fluctuates Over Time
Ratio 2008 Divided by 2002—2% in 50 Years PGA
Ratio 2014 Divided by 2008—2% in 50 Years PGA
Model Volatility Is Largely Driven by “Noise,” and
So Are Your Price Swings if Based on a Model
- 50%
+ 50%
AAL = $5,432.15 $2,716.08
AAL = $4,133.86 $6,200.79
Dispelling Common Myths Surrounding Catastrophe Models
•
The models are not getting more accurate over time.
– Not enough reliable data for any degree of accuracy.
– Much of the volatility in loss estimates results from scientific “unknowns” versus
new scientific knowledge.
– Models are less accurate at finer resolutions. Model updates exhibit the most
volatility for small portfolios with a small number of locations.
•
An updated model is not necessarily a better, more credible model.
– Overspecification combined with high sensitivity of loss estimates to small
changes in model assumptions.
– Overcalibration to most recent event(s).
•
The catastrophe models are not objective tools.
– Most model assumptions are based on subjective judgments of scientists and
engineers rather than objective data.
– Different scientists have their own opinions and biases.
– Scientists can change their minds.
Catastrophe Risk Management Does Not Equal
Catastrophe Models
• Models are a one-size-fits-all approach that cannot be customized to a specific
book of business.
• Risk drivers and risk calculations are not transparent, so it is difficult to test for
credibility and model error.
• Model loss estimates are highly volatile and subject to large swings between
models and model updates:
– Disruptive to underwriting and business strategies.
– Can lead to credibility issues with clients and underwriters.
– Cannot monitor effectiveness of risk management strategies.
• Successful companies employ multiple tools to derive their view of risk.
– Local site conditions and building characteristics can be better indicators of risk.
– Compare model output with available historical information.
– Ability to understand and visualize potential scenarios is critical to understanding
the risk.
What’s the RiskInsight® Open Global Platform?
• Robust, fully transparent and customizable catastrophe loss modeling platform
– Hazard
– Vulnerability
– Financial loss
• Built-in reference models by peril region
–
–
–
–
–
–
U.S. hurricane
U.S. earthquake
Storm-surge flooding
Severe thunderstorm
European windstorm
Japan typhoon
• Deployable in various ways
– Traditional client/server
– Virtual machine
– In your cloud and KCC cloud
What’s a Reference Model?
•
Catastrophe model developed by KCC and external experts
• All events, intensities and damage functions fully transparent
• All components properly validated and peer reviewed
•
Components and assumptions directly customizable by user
Events Are Generated That Are Consistent With
Historical Observations and Latest Scientific Opinion
NE
1 2 3 4 5
Mid-Atl
1 2 3 4 5
SE
1 2 3 4 5
FLNE
FLNW
TX
Gulf
1 2 3 4 5
Relative Wind Speed
High
1 2 3 4 5
1 2 3 4 5
FLSO
1 2 3 4 5
1 2 3 4 5
Low
Full Transparency: 100-Year Texas
Characteristic Events (CEs)
•
Maximum over-land wind speed is
167 mph (Category 5 hurricane).
•
Typical track for region.
Full Transparency: 100-Year Florida CEs
Florida NW
Max. over-land wind
speed varies from 135
mph to 164 mph.
Florida S
Max. over-land wind
speed is 167 mph.
Storm track varies within each region.
Florida NE
Max. over-land wind
speed varies from 135
mph to 164 mph.
PMLs Mask Exposure Concentrations and
Potential Solvency-Impairing Events
100-Year CE Profile
100-Year PML
RiskInsight Illustrates Losses and
Market Shares for Each Event
RiskInsight Produces the Traditional Metrics
•
CE results can be used to produce:
–
–
–
–
Event loss table (ELT)
Year loss table (YLT)
Average annual losses (AALs)
EP curves
Average Annual Loss Costs
Year Loss Table
Differences From Catastrophe Models:
Defined Probability Versus Randomly Generated Events
Historical earthquake data from USGS
Catastrophe Models—Random Events
Fault
location
Magnitude
Random Event 1
Magnitude = 6
Fault Length = 25 km
…..
Fault length
CEs—Defined-Probability Events
Fault dip
Random Event 2
Magnitude = 7.5
Fault Length = 200 km
…..
Characteristic Event 1, 2, ….
Magnitude = 7.1
Fault Length = 80 km
…..
Events are generated by random
sampling from parametric distributions.
Events are generated by identifying the
characteristics with a specific return
period.
Random Event Sampling Compared With
Uniform Geographic Sampling
Random Events at 500-Year Return Period
Sample A
Sample B
CE Methodology
CE-Based AALs Provide Stable and Intuitive View of Risk
CE Loss Cost
Traditional Vendor
Loss Cost
How Companies Are Gaining an Edge
in Catastrophe Risk Management
• Insurers and reinsurers are adopting new tools to address limitations
in traditional catastrophe models:
–
–
–
–
Sampling error
Implementation error
Lack of transparency
Inefficient processes
• Companies pursue competitive advantage by moving beyond the
default models to construct their proprietary view of risk:
– Focus on addressing known deficiencies and leveraging internal
knowledge
– Incremental process
– New tools including RiskInsight are used to build own models
Summary
• Traditional models are useful tools for evaluating catastrophe risk but have
limitations:
–
–
Uncertainty is driven by paucity of data.
Much of the volatility in loss estimates results from noise.
• Dispelling common myths about models:
–
–
–
Models might not increase in accuracy over time.
An updated model is not necessarily more credible.
Models are not purely objective tools.
• Companies are adopting new tools and methodologies to address deficiencies in
traditional models:
–
–
Building proprietary views.
New techniques for increased stability.
• Benefits of CE approach
–
–
–
–
Transparent approach provides deeper insight into catastrophe risk.
Clearly identifies exposure concentrations and “hot spots” that can be missed by PMLs.
More visual and intuitive information for the board and CEO.
Consistent risk metrics for improved portfolio management and risk selection.
Questions?
Download