Transcript Lloyd`s realistic disaster scenarios

Lloyd’s
Realistic Disaster Scenarios
14th October 2004
Background to Lloyd’s

In existence since 1688

World's leading insurance market

66 syndicates

Over 160 brokers

Capacity of approximately £15.0 billion
Profile of the Lloyd’s Market
Lloyd’s
Franchisor
CLIENTS
Reinsureds
Commercial
MEMBERS
168
Lloyd’s
Brokers
66 Syndicates
£15.0bn Capacity
AM Best: A
AS&P: A
Personal
BUSINESS FLOW
Source: Lloyd’s
53
Corporate
Groups
45 Managing Agents
Service
Companies
4
Members’
Agents
2,048
Individuals
Members
(Unlimited
liability)
637
Conversions
(Limited
liability)
CAPITAL PROVISION
Profile of Lloyd’s Capacity & Managing Agents
Ca pa c i t y M ov e me nt
Market Capacit y
£ Bn
16
14
12
10
8
6
4
2
0
No. of syndicat es
No. of Managing Agent
200
150
164
66
100
50
68
45
0
1997
1998
1999
2000
2001
2002
2003
2004
Lloyd’s Realistic Disaster Scenarios
(RDSs)


Lloyd’s RDSs were introduced in 1995

Requiring syndicates to carry out disaster planning

Based on a range of hypothetical scenarios
Scenarios mainly based on historical disasters

“Failure of North Sea Complex” – from Piper Alpha

“Marine Collision” – from Exxon Valdez

“Japanese Earthquake” – from 1923 Great Kanto Event
RDS Overhaul Project



Two year RDS Overhaul Project

Improve consistency of risk assessment

Raise Loss Modelling standards in the market
Collaborative Process

Market Experts Group

Lloyd’s Market Association – Marine RDS

AIR, EQECAT and RMS

Florida Windstorm

California Earthquake
Overhaul to be Completed in April 2005

Will take account of feedback received in 2004
Why Overhaul the RDSs?

To provide a more reliable and accurate measurement of risk,
based on a more comprehensive and consistent approach


To measure the market’s and syndicates’ exposures to a series of
major catastrophes

RDS assumptions and practices varied widely

Significant variations observed in “Aggregate” to “Gross Loss”
relationships

Wide variations in methodologies
To provide a better understanding of Lloyd’s aggregate risk profile

For catastrophes

As an input to capital setting and allocation decisions

As an input in managing Central Fund exposures
RDS Guidance and Instruction

Issued in April 2004



Replaces previous “Regulatory Bulletins”
with a comprehensive guide to the
assessment of catastrophe exposures
REALISTIC DISASTER SCENARIOS APRIL 2004
Key sections include

Definitions

Calculation Principles

Recommended Best Practice

“Event Subset” Submissions

Scenarios and Events
Available on Risk Management Website

www.lloyds.com/riskmanagement
Guidance and Instructions
Scenarios and Events
SINGLE RISK SCENARIOS
AGGREGATING SCENARIOS
1
USA Windstorm
11
Second Event
2
Marine Event
12
Florida Windstorm
3
North Sea – Loss of Major Complex
13
California Earthquake
4
Aviation Collision
14
New Madrid Earthquake
5
Major Risk Loss
15
European Windstorm
6
Satellite Risks
16
Japanese Earthquake
7
Liability Risks
17
Terrorism
8
Political Risks
9
Alternative RDS: A
10
Alternative RDS: B
Marine RDS


Syndicates must return two scenarios

Collision in Prince William sound

Major Cruise Vessel Incident
Increased guidance for both scenarios

Specification of losses to the International Group of P&I Associations
Reinsurance Programme.
Marine collision in Prince William Sound
Cruise Vessel
Tanker Owner
Protection and Indemnity
Protection and Indemnity
US$ 2,050m
US$ 2,050m
Layer 4
Layer 4
US$ 1,550m
US$ 1,550m
Layer 3
Oil Pollution
US$ 1,050m
Oil Pollution
Layer 2
Layer 2
US$ 1,050m
Layer 2
Layer 2
POOL RETENTION
Owned Entries
Layer 1
US$ 187.5m
US$ 50m
Layer 1
POOL RETENTION
Owned Entries
Coinsurance 25%
Layer 1
Coinsurance 25%
Layer 1
Coinsurance 25%
US$ 437.5m
Coinsurance 25%
US$ 600m
US$ 550m
US$ 550m
US$ 50m
Layer 3
California Earthquake
EXPOSURE DISTRIBUTION
EVENT FOOTPRINTS
Catastrophe Loss Modelling Software –
Florida Windstorm & California Earthquake

Three agencies - AIR, EQECAT & RMS

Working with the three catastrophe modelling agencies has
enabled us to:

Specify new events tightly, by industry loss level

US$70bn and $54bn respectively


Specify 2 events of similar magnitude


Based on Commercial (including Industrial) and Residential losses
Better coverage of risk (e.g. LA and San Francisco)
Derive Average Damage Factors

Feasible events cannot destroy all properties
Florida Windstorm and
California Earthquake

County level “Footprints”


County level damage factors


Derived from discussions with AIR, EQECAT and RMS
Prompt for additional lines of business to include


“Aggregate” to be based on county exposures in “Footprint”
Marine, Specie/Fine Art, Personal Accident, Aviation, Liability
Exclusion of classes viewed as “impossible” to model

Contingent Business Interruption

Aviation Hull in Earthquake
RMS “Roll-up” exercise

Pilot study with five managing agents


Combining Event Loss Table output

To validate choice of new RDS events

To assess feasibility of combining syndicate results
Analysis split by broad business categories

Direct & Facultative

Treaty

Binder
RMS “Roll-up” exercise
– California EQ
D&F
Lloyd's Sample Loss
D&F Benchmark Events
RMS Industry Loss
RMS “Roll-up” exercise
TREATY
Lloyd's Sample Loss
Treaty Benchmark Events
RMS Industry Loss
RMS “Roll-up” exercise
BINDERS
Lloyd's Sample Loss
Binder Benchmark Events
RMS Industry Loss
Terrorism RDS
Terrorism RDS


2-tonne conventional bomb attack on Empire State Building

Damage zones and levels specified in scenario

TRIA and non-TRIA events

Specification of deaths, injuries and BI

Fire-following, Number of Events and CBRN Guidance
Focus on data resolution

Address-level best for urban accumulations

Guidance for proportions in 10001, 10016 and 10018
Terrorism RDS
Data Resolution
Source: Munich Re. Topics Geo 2003
RDS Results
Playback

Visits

Each Managing Agent visited by two or more Risk
Executives

General questions asked concerning:


Data capture and resolution of data

Identifying relevant policies

Loss calculation and expected values

The return process & quality assurance

Modelling methods
Review of one RDS scenario from start to finish
RDS Results
Playback

Packs

Every syndicate receives a pack prepared solely for their own use.

Contains syndicate specific analysis on:

Aggregate, Gross, Net and Final Net losses

Losses in relation to capacity and hence Franchise Performance
Guidelines

Ranking within the market

Includes analysis on modelling types at both market and syndicate
level for each scenario.

Graphical representation of Losses, Average Damage Factors
(ADFs) and Ranking.
RDS Results
Modelling Type Analysis
Modelling type - Market
Licensed Softw are(AIR/EQECAT/RMS)
Modelled Internally
Maximum Line
Market Share
Premium Derived
M arine Co llisio n in P rince William So und
Sunk/Damaged Cruise Liner
M ajo r Risk Lo ss
Seco nd Event
USA Windsto rm
A lternative RDS B
No rth Sea Co mplex
A lternative RDS A
Specific Euro pean Windsto rm
Lo s A ngeles
San Francisco
Specific New M adrid Earthquake
Specific Japanese Earthquake
M iami
Tampa B ay
Liability
No n-TRIA Event
TRIA Event
0%
10%
20%
30%
40%
50%
60%
70%
Percentage of Gross Loss by Modelling type
80%
90%
100%
Four-Stormed is Forearmed
14 October 2004
How Hurricanes are Formed

Thunderstorms drift over warm ocean
waters.

Warm air from the storm and ocean
surface combine and begin rising,
creating low pressure.

Trade winds blowing in opposing
directions cause the storm to start
spinning.

The rising warm air causes pressure to
decrease at higher altitudes.

Air rises faster and faster to fill this low
pressure, in turn drawing more warm
air off the sea and sucking cooler, drier
air downwards.

As the storm moves over the ocean it
picks up more warm, moist air. Wind
speeds start to increase as more air is
sucked into the low pressure centre.

It can take hours or several days for a
depression to grow into a fully-formed
hurricane.
Lloyd’s 2004 Realistic Disaster Scenarios
Distribution of Florida Property Values
Hurricane Charley – The facts

Hurricane Charley made landfall at Sanibel Island,
Florida, late on the afternoon of 13 August 2004.

Maximum sustained winds were near 145 mph after
upgrading to a Category 4 storm, causing severe
damage and flooding in coastal areas.

Hurricane Charley had been expected to make landfall
near Tampa, 120 miles to the north, but shifted course
at midday and brushed low-lying islands.

Charley was responsible for the death of 27 people in
Florida.
RDS 2004 Pinellas Hurricane Event Footprint
Estimation of track of
Hurricane Charley
Hurricane Frances – The facts

Hurricane Frances made landfall at Sewall’s Point in
Martin County, Florida, at 05:00 GMT on Sunday 5th
September.

Maximum sustained winds were near 105 mph and
Frances dropped more than 33cm of rain along
Florida’s east-central coast.

Insured losses will be limited because much of the
damage has been caused by flooding, which is
covered by the National Flood Insurance Program.

Frances was responsible for the death of 23 people in
Florida.
RDS 2004
Miami-Dade Hurricane Event Footprint
Estimation of track of
Hurricane Frances
Tracks of Hurricane Charley & Hurricane Frances
Source: Guy
Carp.com.
Cat-i Reports
Hurricane Frances
Update- Report 10,
September 7, 2004
Hurricane Ivan – The facts

Hurricane Ivan made landfall just west of Gulf Shores in
Baldwin County, Alabama at 07:00 GMT on Thursday 16th
September.

Maximum sustained winds were near 130 mph when it
came ashore and subsequently moved inland and
weakened.

Ivan was responsible for the death of at least 50 people in
the United States in addition to over 70 people in the
Caribbean.
Tracks of Hurricanes Charley, Frances & Ivan
Source: Guy
Carp.com.
Cat-i Reports
Hurricane Ivan
Update- Report 13,
September 17, 2004
Hurricane Jeanne – The facts

Hurricane Jeanne made landfall near Hutchinson Island
just east of Stuart in Martin County, Florida at 04:00 GMT
on Sunday 26th September.

Maximum sustained winds were near 120 mph (Cat 3) when
it came ashore and subsequently moved inland and
weakened.

Jeanne was responsible for the death of at least 3,000
people in Haiti mainly due to flooding and 8 people in
Florida.
Track of Hurricane Jeanne
Source: Guy
Carp.com.
Cat-i Reports
Hurricane Jeanne
Update- Report 7,
September 27, 2004
Tracks of the Four Hurricanes
Source: Guy
Carp.com.
Cat-i Reports
Hurricane Jeanne
Update- Report 7,
September 27, 2004
Industry Loss Estimates
RMS
EQECAT
AIR
Charley
USD 6-8bn
USD 5-10bn
USD 6-10bn
Frances
USD 3-6bn
USD 2-5bn
USD 5.6-8.7bn
Ivan
USD 3-6bn
USD 4-10bn
USD 3-6bn
Jeanne
USD 4-8bn
USD 6-14bn
USD 5-9bn