Transcript An Overview to Client name - The University of the West Indies at

Exposure,hazard,vulnerability – insurance risk management with and without
climate change
Mr. Andrew Mitchell
1
©Copyright 2007 Willis Limited all rights reserved.
exposure, hazard, vulnerability –
insurance risk management
with and without climate change
Andrew Mitchell
Willis Analytics
Climate Change: Impacts on the Caribbean
Caribbean Community Climate Change Centre
University of the West Indies
16 June 2007
insurance risk identification and quantification
•
the cost-benefit of insurance - smoothing, pooling, diversifying
•
experience-based rating - generalised linear models
•
poor identification of extremes - no geography, no science, no engineering
•
reinsurance provides cover for catastrophe event accumulations of loss
•
cat modelling introduced to reinsurance sector in early 1990s
•
discipline this created is as important as the numbers in output
•
cat modelling has become a pre-requisite for new capital, esp Bermuda
•
provided demonstrable stability - far fewer failures in 2000-2005 vs 1990-1995
•
cat modelling is cross-disciplinary and multi-component
•
stochastic event sets
•
probabilistic treatment of primary and secondary uncertainty
3
catastrophe loss models: structure
HAZARD
VULNERABILITY
FINANCIAL
• event generation
• mean damage ratios
• local intensity calculation
• building inventory
• applying insurance terms
and conditions
location details
policy conditions
EXPOSURE
• sums insured
• risk type
• coverage type
4
catastrophe loss models: output
Island
Parish
Occupancy
Construction
Height
Yearbuilt
Return Period
1000
750
500
250
200
100
50
25
10
5
Jamaica
St. James
Temporary Lodging
Unknown
Unknown
Unknown
Occupancy only
61.36%
56.49%
49.48%
37.50%
33.74%
22.59%
12.72%
5.00%
0.23%
0.00%
Jamaica
St. James
Temporary Lodging
Reinforced Concrete
Unknown
Unknown
+ Construction
57.41%
52.72%
46.06%
34.82%
31.29%
20.84%
11.58%
4.40%
0.17%
0.00%
Jamaica
St. James
Temporary Lodging
Reinforced Concrete
4 stories
Unknown
+ Height
46.01%
41.26%
34.65%
23.87%
20.60%
11.40%
4.39%
0.69%
0.00%
0.00%
Jamaica
St. James
Temporary Lodging
Reinforced Concrete
4 stories
2006
+ Year Built
44.17%
39.53%
33.11%
22.64%
19.47%
10.60%
3.95%
0.57%
0.00%
0.00%
80%
Occupancy only
+ Construction
+ Height
+ Year Built
70%
gross loss OEP as % of insured value
60%
model sensitivity analysis
RMS RiskLink v6.0
hurricane EP curve
primary modifier variations
50%
40%
30%
20%
10%
0%
0
100
200
300
400
500
return period (year)
5
600
700
800
900
1,000
exposure: US land-falling hurricane losses – actual
source: Normalized Hurricane Damages in the United States: 1900-2005
Pielke et al Natural Hazards Review (submitted)
6
exposure: US land-falling hurricane losses – revalued
source: Normalized Hurricane Damages in the United States: 1900-2005
Pielke et al Natural Hazards Review (submitted)
7
exposure: growth and data capture
•
population growth
•
growth in property values
•
growth of urban concentrations
•
settlement and development in exposed regions
•
rise in standard of living
•
increased international trade - marine cargo exposure
•
increased insurance penetration
•
increased correlation means exposure to cat events rises faster than income base
•
insurance exposure data:
• capture and reporting limited by legacy systems
• quality and type has not been standardised
• trans- and multi-national policies mean location identification is confused
• ? largest source of error in modelling
8
hazard: source of losses
hurricane
earthquake
hurricane
earthquake
9
hazard: relative size of losses
hurricane
earthquake
RMS v6.0 Jamaica earthquake - buildings only
25.00%
25.00%
20.00%
20.00%
OEP gross loss as % of total insured value
OEP gross loss as % of total insured value
RMS v6.0 Jamaica hurricane (long-term historical event set) - buildings only
15.00%
10.00%
IED WS (LT)
50:50 Res:Com WS (LT)
20:80 Res:Com WS (LT)
20:60:20 Res:Com:Hotel WS (LT)
5.00%
15.00%
10.00%
IED EQ
50:50 Res:Com EQ
20:80 Res:Com EQ
20:60:20 Res:Com:Hotel EQ
5.00%
0.00%
0.00%
0
100
200
300
400
500
600
700
800
900
0
1,000
100
200
300
400
500
600
return period (years)
return period (years)
other types of hazard:
other classes of business:
•
fire
•
liability
•
theft
•
health
•
explosion
•
aerospace
•
terrorism
•
marine
•
volcanic eruption
•
credit
•
finex
10
700
800
900
1,000
hazard: complications
•
secondary hazards not modelled - landslip, fire following earthquake
•
other non-modelled loss contributors - economic loss, vulnerability factors
•
impact of climate change will vary geographically and over time
•
existing hazards change, new hazards emerge
•
impact on frequency and severity unclear
•
data too limited:
• current trends vs natural variability
• local uncertainty - confused by feedback systems
• downscaling and regional modelling not ready
•
climate change - just another source of uncertainty?
11
vulnerability: other components of insurance loss
•
mean damage ratios
• engineering - design and construction components
• calibrated by loss and claims experience
•
non-linear factors
• economic demand surge
• claims inflation - fraud, propensity to claim
• business interruption
• loss adjusters expenses
•
claims management strategy
• approved suppliers, approved repairers
• regional restrictions
• saturation point unclear
• poor repair quality vs retro-fitting
•
natural increase in vulnerability and inter-dependence of systems
•
political interference
12
insurance: risk management and climate change
•
growing connection between insurance and academia
•
contextual information vs “plug-in” data
•
the flat earth has returned - data and modelling required for everywhere
•
solvency and regulation - risk-based capital
• underwriting, reserving, credit, operational, liquidity, investment
•
insurance as economic not social function
•
reduction in insurance relief relative to economic losses
•
risk appetite to meet return-on-capital targets
•
interface between insurance and public policy
•
non-stationarity means limited insurance adaptation is possible
•
insurance cannot provide cover for the predictable
13