Resilience in Insurance - Centre for Science and Policy

University of Cambridge
Centre for Science and Policy
8th March, 2011
Resilience in Insurance
Stress test business plans for extreme events – hope for
the best; plan for the worst
Dr Dougal Goodman FREng
Chairman
Lighthill Risk Network
[email protected]
The Lighthill Risk Network is a consortium of insurance
companies and brokers working to build bridges between the
insurance market and the research community.
Useful URLs
•
•
•
•
•
•
Blackett Review on High Impact/Low Probability Risks
www.bis.gov.uk/assets/bispartners/goscience/docs/b/12-519-blackettreview-high-impact-low-probability-risks.pdf
The Foundation for Science and Technology
www.foundation.org.uk
The Risk Centre, Cranfield University
www.cranfield.ac.uk/sas/risk
Professor Richard Smith
www.stat.unc.edu/faculty/rsmith.html
Professsor Paul Embrechts
www.math.ethz.ch/~embrechts/
Cambridge Centre for Risk Studies
www.risk.jps.cam.ac.uk
Dr Dougal Goodman, Lighthill Risk Network
1
Paul Thompson
Dr Dougal Goodman, Lighthill Risk Network
Paul Thompson
Dr Dougal Goodman, Lighthill Risk Network
2
Back up propulsion
Capability of the
Crew
Weather forecast
Skipper’s Risk
Judgement
Currents and tides
Emergency
preparedness
Strength of the hull
and keel
Travel safely in the
shortest time from
St Malo to Cowes
Wind speed and
sails
Water depth
Quality of charts
Local knowledge
Routing Information
Dr Dougal Goodman, Lighthill Risk Network
Risks to a Company
Dr Dougal Goodman, Lighthill Risk Network
3
System of Systems
Global Market
Regions
Markets and
Regulation
Sectors
Company
Threat of contagion
Dr Dougal Goodman, Lighthill Risk Network
Events do happen…
FTSE 100 Index
8000
7000
6000
Index
5000
4000
3000
2000
1000
0
01 January
1984
01 January
1988
01 January
1992
01 January
1996
01 January
2000
01 January
2004
01 January
2008
01 January
2012
Year
2007 Floods
Banking Crisis
Japanese Tsunami
Eurozone Crisis
What next?
Dr Dougal Goodman, Lighthill Risk Network
4
Deep downsides – only some can be quantified
• Sudden product price or • Business interruption
•
•
•
•
•
•
margin fall
Customers do not pay
for goods or services
Liability claim
Loss of trust of
employees
Outrage of the public
Projects fail to complete
on time and within
budget
Cyber attack
•
•
•
•
•
•
from natural
catastrophes
Fire or explosion at a
facility
Fraud, accounting or
trading failure
Collapse of a supplier
Regulatory or tax step
change
Loss of utilities
and so on…
Risk level controlled by elimination, management or transfer
Dr Dougal Goodman, Lighthill Risk Network
What is acceptable?
A board decision
As Low As Reasonably Practicable
Risk = Consequences * Vunerability * Likelihood
Risk = How big? * How vunerable? * How often?
Impact per unit time
Dr Dougal Goodman, Lighthill Risk Network
5
What could destroy 5%
or more of UK GDP or
5% or more of the
value of a company?
Dr Dougal Goodman, Lighthill Risk Network
Macondo Well Failure
• Damage to BP brand in US and elsewhere
• Combination of events led to low probability,
high severity event
• Challenge to client contractor relationships
• Significant loss of value to the company
Dr Dougal Goodman, Lighthill Risk Network
6
Macondo Insurance Exposure
Total economic loss:
$40 billion plus
Working
interest
Maximum
insurance coverage
BP
65%
Self-insured
Anadarko
25%
$163 million
Mitsui
10%
$175 million
Transocean
Contractor
$950 million
Dr Dougal Goodman, Lighthill Risk Network
Lloyd’s of London
commitment to insureds
is to “settle all valid
claims”
Market made up of
underwriting syndicates
regulated by Lloyd’s
corporate structure
Dr Dougal Goodman, Lighthill Risk Network
7
Risks to an insurance
syndicate
Dr Dougal Goodman, Lighthill Risk Network
Solvency II – Regulation of Insurers
New regime for the regulation of insurers from January, 2014
Pillar I -
Companies will have to show regulators that they have
appropriate internal models for evaluating their capital
requirements or use a standardised approach
Pillar II –
Demonstrate that they have an effective risk
management process and a risk identification process
Pillar III –
Public disclosure and reporting requirements
Test of the ability of an insurance entity to survive a 1 in 200 event.
Regulators wish to be reassured that if premiums are collected the
insurer will be able to pay valid claims for future losses.
Dr Dougal Goodman, Lighthill Risk Network
8
Typical time series of losses for a catastrophe line of business
Typical Year
40
30
20
10
0
01-Jan
01-Jul
Individual Events
100
80
60
40
20
0
1990
1995
2000
2005
2010
Aggregate Annual Totals before Recoveries
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
100
80
60
40
20
0
Hypothetical data
Dr Dougal Goodman, Lighthill Risk Network
Partial Correlation
Flood
Quake
Energy
UK, Europe,
SE Asia
California,
Iceland, New
Zealand
North Sea, Gulf
of Mexico, Brazil
Windstorm
Marine
NW Europe,
Florida, Gulf
Coast, SE Asia
Tankers, Cruise
Ships, Bulk
Carriers, Box
Ships
Typical Underwriting Book of Business
Dr Dougal Goodman, Lighthill Risk Network
9
Reinsurance Premiums
Reinsurance Recoveries
Premiums
Underwriting Book
Claims Paid and
Estimated
Administration
Costs
Investment Portfolio
FREE RESERVE
Liabilities
Reserves
FREE RESERVE
90
80
70
Dividends and Interest
60
50
40
30
20
10
0
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
Dr Dougal Goodman, Lighthill Risk Network
Extreme Value
Analysis
Dr Dougal Goodman, Lighthill Risk Network
10
Compulsory Scenarios
Two Events
Florida Windstorm
Gulf of Mexico Windstorm
European Windstorm
Japanese Windstorm
Californian Earthquake
New Madrid Earthquake
Japanese Earthquake
UK Flood
Terrorism
Optional Scenarios
Marine P&I Risks
Loss of Major Complex
Aviation Collision
Satellite Risks
Liability Risks
Political Risks
Alternative RDS
Dr Dougal Goodman, Lighthill Risk Network
UK Flood Event – Lloyd’s RDS Scenario 9
Residential
£4.5 bn
Commercial/industrial
£1.6 bn
Agriculture
£0.05 bn
Motor
£0.05 bn
TOTAL
£6.2 bn
To estimate the loss a
knowledge of precipitation,
surface flooding, river flow,
terrain heights, land use and
the social aspects of how
people react to flood warnings
is required.
Dr Dougal Goodman, Lighthill Risk Network
11
Risk
Appetite
Realistic
Disaster
Scenarios
acceptable free
reserve volatility
Historical
Loss
Analysis
Agreed
Insurance
Strategy
Model
Reinsurance
Programme
Options
Trends
Factor
Analysis
for future
losses
The model
cannot make
the decision
Market
Review
Dr Dougal Goodman, Lighthill Risk Network
The Tail
Generalised Pareto
Mean of 5, standard deviation 1
Shape parameter is 0.6, beta is 1
0.2
0.1
0.2
0.4
0.3
0.6
0.4
Gaussian
0
2
4
0.0
0.0
3σ
6
8
10
0
2
4
6
8
10
x
Break out of thinking that all distributions are normal distributions.
Tail events may, however, be driven by a different process.
For extreme events it can usually be assumed that they are
uncorrelated and follow a Poisson Process
Dr Dougal Goodman, Lighthill Risk Network
12
Communication of Risk
Dr Dougal Goodman, Lighthill Risk Network
Dimensions for Communication
• Economic Impact
Impact on the economy or business return
• Human Impact
Fatalities and serious injuries
• Outrage
Lost of trust in government or business management
• Social and Environmental Disruption
Social breakdown or severe environmental damage
• Likelihood
Return period for an event – large uncertainty
Depends on scenario description
Dr Dougal Goodman, Lighthill Risk Network
13
Unbundling Risk
Transport Disaster
Economic
Impact
Outrage
Transport Disaster
Social and Environmental Impact
Human Impact
The area of the circle is a measure of likelihood
Dr Dougal Goodman, Lighthill Risk Network
Risk Framework of Macro-Threats
Dr Andrew Coburn, Cambridge Risk Studies Centre
1
Financial
Shock
5
Natural
Catastrophe
9
Disease
Outbreak
2
Trade
Dispute
6
Climatic
Catastrophe
10
Humanitarian
Crisis
3
Geopolitical
Conflicts
7
Environmental
Catastrophe
11
Externality
4
Political
Violence
8
Technological
Catastrophe
12
Other
Shock
Community peer review being conducted at http://systemshock.org.uk
Dr Dougal Goodman, Lighthill Risk Network
28
14
Analytical
Emotional
Engineers
Officials
Politicians
Scientists
Economists
Analysts
Managers
Lawyers
Intermediaries are needed to translate….
Dr Dougal Goodman, Lighthill Risk Network
Summary
• Every organisation should have systems in place to
test the impact of deep downsides on future
performance
• Senior managers should stand back from day to day
operations and think about the events that could
destroy the company or organisation
• Communication of extreme value analysis is not
straightforward – a new language for communicating
risk in business is required
Hope for the best; plan for the worst
Dr Dougal Goodman, Lighthill Risk Network
15
Thank You
Dr Dougal Goodman, Lighthill Risk Network
Dr Dougal Goodman, Lighthill Risk Network
16