Natural Hazard Property Losses & Climate Change
John McAneneyRisk Frontiers
Macquarie UniversitySydney NSW
Overview
• Risk Frontiers• Risk Assessment• Climate Change • North Atlantic Hurricane Windspeed Data• Normalisation of ICA loss database• Property losses from bushfires• Policy implications• Natural Hazard Risk Profiles
Risk Frontiers
An independent and local research capacity to help insurers better understand and price natural hazard risks in the Asia-Pacific Region by:
• Undertaking research in natural hazards• Undertaking post-event reconnaissance• Developing Probabilistic Catastrophe Loss models• Increasing public awareness of natural perils
Conceptual framework of risk assessment
Hazard
Exposure
Vulnerability
Risk
Risk = f (Hazard, Exposure, Vulnerability)
Mean intensity
Mean d
am
age r
ati
o (
%)
Loss ($ Million)
Annual Exce
edance
Pro
babili
ty
Nicene Creed of Climate Change
• Global mean and extreme temperatures are increasing
• Heating is due to increasing atmospheric CO2 and other GH gases
• Warming is occurring where models suggest it should under increasing CO2
• Sea level is rising• Take greenhouse gases out of the models, earth cooling slightly• If reject increasing GH gases as explanation, need to find some
other hypothesis
Uncertainties
• GCMs are too complex to be fully understood and the climate system depends upon many ingredients that must be represented either empirically or through ad hoc treatments that differ between models
• Arguments over the scale and speed of warming in the future
• Models have little to say yet about regional implications
• Models can’t resolve phenomena like droughts, floods, storms, cyclones, etc.
• Attribution of individual weather events to climate change
Costs of weather-related natural catastrophe losses are increasing: why?
Source: Swiss Re sigma Catastrophe database
1992:Hurricane Andrew,
USD 22 bn
1999:Storms
Lothar/Martin,USD 10 bn
2004:Hurricanes Charley,
Frances, Ivan, Jeanne,
USD 29 bn
2005:Hurricanes Katrina,
Rita, Wilma,USD 65 bn
USD
mill
ions
(20
05
$
)
Distribution by hazard worldwide of the largest 40 insured losses
1970-2004(Source: Swiss Re)
Atlantic basin hurricane data – Wind speed
Atlantic basin hurricane tracks (Category 1-5) during 1851-2006
Difference of wind speed distributions between the early historical period (1851-1946) and the recent six decades.
- Early historical records significantly underestimate the frequency of Category 4-5 winds.
- Wind speed distributions over the past two, four and six decades display little systematic changes.
U.S. landfalling segments since 1947
Damage to property and other assets is linked to landfalling events.
For the six decades since 1947, there are no sustained upward trends in - Average annual count of landfalling segments (blue curve) - Mean landfalling wind speeds (red curve)
This contrasts with the dramatic increases in total economic and insured losses, suggesting the losses must be attributed to factors other than wind speed alone.
Australian Property Losses over 20th Century
Australian Losses - ICA Natural Disaster Event List
• Insurance Council of Australia database of insured losses since 1967
• Estimate losses as if events took place in 2006
• Account for changes in – Inflation– Population– Wealth– Building Codes
Original Annual Aggregate Losses (July 1 – June 30)
Original annual aggregate insured losses (AUD$ million) for weather-related events in the ICA Disaster List for years beginning 1 July
Methodology
CL06 = Li × Ni,j × Di,k x Btc
CL06 - Normalised (current) dollar loss (year 2006 value)
Li - Original dollar loss (year ‘i’)
Ni,j - Dwelling number factor
Di,k - Dwelling value factor j - Urban Centre / Locality (UC/L) impacted by the event k - State or Territory that contains the impacted UC/L Btc - Building Code adjustment (= 1 for all hazards except tropical cyclones)
Number of occupied dwellings in the Sydney UCL
600000
700000
800000
900000
1000000
1100000
1200000
1300000
1400000
1966 1971 1976 1981 1986 1991 1996 2001 2006
Year
Nu
mb
er
of
Oc
cu
pie
d D
we
llin
gs
Average nominal new dwelling value (AUD$ thousands) for WA.
0
20
40
60
80
100
120
140
160
180
200
1967 1972 1977 1982 1987 1992 1997 2002
Year
Ave
rag
e N
om
inal
Val
ue
(AU
D$
tho
usa
nd
s)
Building Code Adjustment
• Estimate % of loss due to wind vis-à-vis flood • Proportions of pre- & post-1981 construction
then and now• Use Central Pressure at landfall to determine
characteristic gust speed for the cyclone• Calculate pre- & post-1981 loss ratios• Adjust normalised loss• Unique adjustment for each event
Attributes
• Uses publicly-available information• Based on dwellings rather than population
– Number of dwellings ~ Population– Nominal value of new dwellings ~ Inflation &
Wealth
• Nominal dwelling value excludes land value– Assures reasonable alignment to insured losses
• Easy to apply
TC Tracy - 1974• Original loss = $200M
– Dwelling number factor ~ 3– Nominal dwelling value factor ~13
• (1974: ~$18.5K; 2006: ~ $240K)
– All losses attributed to wind or wind-driven rain– Current construction all post-1981– Building code factor ~ 0.5
• Current loss ~ $3.6 billion• This may be slightly high as building code regulations
were introduced in Darwin earlier than 1981
Normalised Annual Aggregate Losses (July 1 – June 30)
Brisbane Flood
TC Tracy
Ash Wednesday Fires
Sydney Flood & Brisbane Hail
Sydney Hail
Sydney Hail
Top 10
Attribution of Loss
Annual Australian Bushfire Losses
Bushfire loss frequency
Time Series Analysis to 2003
A “major” event is defined here as more than 25 homes destroyed
within a 7 day period.
Hurricane Damage if landfall in 2005
$-
$20
$40
$60
$80
$100
$120
$140
$160
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
Bill
ion
s
Year
Lo
sses
in 2
005
US
D
Population Changes
Florida Coastal Population1900 to 1990
Year 15 million today!
Implications for Insurers
• Societal factors predominant drivers of increased natural disaster losses
• No need to invoke global climate change for increasing losses – not yet
• Expect this to be the case over the next few decades
• Insurers need to worried about what might happen in the next twelve months or so
Public Policy Implications
• Efforts to reduce society’s vulnerability to current & future extremes
• Improved wind standards best example
• Bushfire: restrictions based on distance to forest
• Flood: limit construction on floodplains
• Risk reduction (adaptation) measures in addition to abatement of greenhouse gases
Looking X years ahead
• Ability to obtain insurance linked to the actual risk
• Premium will be linked to actual risk– distance from bushland?– are you on a floodplain?– distance from the coast?