earthquake model
The use of catastrophe models in the TCIP and other national insurance schemes Shigeko Tabuchi
6th DPRI-IIASA Forum 15 August 2006, Istanbul, Turkey
Making better reinsurance decisions…
about Willis • A leading global insurance broker - We are one of the world's leading risk management and insurance intermediaries - We are in the business of identifying, analyzing and managing risk - We deliver professional risk transfer, risk management, loss management and actuarial services to companies, as well as financial and employee benefits consulting.
• Company originated in 1828 • We have approximately 15,800 people around the world, with 300 offices in 80 countries • We serve both private and public clients in over 180 countries - clients include; corporations, public entities and institutions
• Willis Analytics & Solutions - The team has 150 staff in the UK, US and a network of specialists - Expertise split between: o Catastrophe Management Services o Development & Support o Financial Management Services o Marketing, Applications and Delivery o Willis Integrated Solutions 2
why model catastrophe risks ?
• Catastrophe (natural perils) risks are the major hazards that insurers are exposed to • earthquake • flood • windstorm
• The cost of natural hazards is continually increasing • Increased urbanisation • More people buy insurance
A high proportion of the population is in the highest risk zone ...and there has been significant seismic activity in the most populated areas in Turkey
3
• Every major city in the world is exposed to one or more natural hazards
why national natural catastrophe pools and programmes? • To lessen the effect of natural catastrophes on public and private finances • To achieve risk distribution and to spread the financial costs of natural catastrophe to international reinsurance and capital markets • For pools, to build up funds within a pool to reduce the requirement for external risk distribution
IDENTIFICATION
QUANTIFICATION
PRIORITISATION
PRIVATISATION
Which hazard exposures? e.g.
Impact of natural hazard
• Actual (current cost)
• Pooling mechanism
• Volcano
• Economic
• Potential cost
• Pricing
• Earthquake
• Government finances
• Recovery potential • Premium collection • Financing options • Distribution • Action plans channels
• Windstorm • Biohazard • Flood
• Social
• Hail
• Claims payment
• Drought
• Mitigation
4
EXPORT
• Structure • Price • Marketing • Leverage • Economic benefit
example of Turkish Catastrophe Insurance Pool (TCIP) • Launch:
2000
• Scope:
residential buildings
• Peril:
earthquake, flood
• Compulsory:
yes
• Cover:
originally US$ 28,000 (2000), currently approx. US$ 70,000. liable to change due to inflation and fluctuations in Turkish New Lira exchange rate
• Administration: premium collected by insurers and scheme administered by the pool manager • Risk transfer:
international reinsurance
• Guarantee:
Turkey Government (via World Bank loans)
Points of interest: • Regarded as a success, and many other countries are interested to follow its example • First project to carry out the whole process from catastrophe modelling to risk transfer for a whole country • First World Bank risk privatisation project • Take up rate for cover is average of C.15% to maximum of C.25% (in Istanbul). • Large capacity for reinsurance programme was available shortly after major earthquake event (Izmit 1999).
5
risk modelling at national level - Turkey Turkish Emergency Flood and Earthquake Recovery project (TEFER) • The World Bank wanted to avoid acting as a quasi-(catastrophe) reinsurer • The catastrophe risk modelling and establishment of an insurance pool were requirements of the World Bank, in order for emergency funds to be released after the Koecaeli earthquake in 1999 • The project was tendered according to strict World Bank rules, with a requirement for a very high “technical score” • Willis Re was appointed by the Turkish Government and the World Bank to create a new earthquake insurance model and design / reinsure a national pool
6
example of Algerian Catastrophe Insurance Programme (ACIP) • Launch:
2004
• Scope:
residential buildings, commercial buildings and contents
• Peril:
earthquake, flood, other natural catastrophe
• Compulsory:
TBD
• Cover:
under the Obligatory Insurance against Natural Catastrophe
• Administration: local insurers provide insurance cover to property owners, • Risk transfer:
Insurers have 30% (retained) to 70% (ceded) quota-share with CCR (government owned, state reinsurer) and CCR purchases reinsurance internationally
• Guarantee:
CCR does not have a reserve fund but Algerian state provides unlimited guarantee to CCR
Points of interest: • Take up rate is not as high as TCIP but slowly picking up • Covers commercial properties • High population and high seismic risk area along the northern coast • Large capacity for reinsurance programme available
7
how do catastrophe models work?
“cat model” loss estimation software
INPUT
OUTPUT
Exposure Exposuredata data
Expected Expectedlosses losses Hazard model
Inventory
Vulnerability functions
8
Financial Models
earthquake model – key inputs Key inputs: • Historical events database Source: KOERI
• Neotectonic data • Seismic source zones • Ground condition data
Active faults
Mathematical relationships:
10.00 PGA (g)
Bommer et al. 2003 Rock
• Fault rupture dimensions • Recurrence relationship - to calculate annual occurrence of an event
1.00
Bommer et al. 2003 Stiff Soil Bommer et al. 2003 Soft Soil
Seismic source zones
Bommer et al. 2003 Stiff Soil + 1s.d. Recorded PGA (geometric mean)
0.10
• Attenuation equation - to calculate ground motion or intensity values at a given location
0.01 1
10
d (km)
Attenuation equation
9
100
1000
vulnerability functions Damage / Loss experience
Engineering approach Demand Curve v Capacity Curve for building type A Spectral Acceleration Sa (g)
2003 Boumerdes earthquake
1.6 1.4 1.2 1.0
Basic Demand Curve Damped Demand Curve
0.8
Capacity Curve 0.6 0.4 0.2 0.0 0.00
Source: CRAAG
0.10
0.20
0.30
0.40
0.50
Spectral Displacement Sd (metres)
RC Lowrise, good construction MeanDamage Damage Ratio Mean Ratio
0.6 0.5
EMS C&S
0.4
CAR:RMS CAR:Willis
0.3
Radius
0.2
FH
0.1 0 5
6
7 8 MMI Intensity
9
Source: CAR
10
10
0.60
exposure information Where is it?
CRESTA Zone
- Location information (country, state/province, district, city, postcode, full address, latitude-longitude…)
What is it? - Risk type (residential, commercial, industrial) - Coverage type (Building, Contents, Business Interruption) - Building construction type, age, height,
Province
How much is it worth? - Property value - Sum Insured and Policy Count - Insurance policy conditions (deductible, limit, coinsurance)
11
District
national building stock dataset • Turkish building classified into 15 buildings types according to their known seismic performance
• Algerian building classified into 15 building types • National census data other official database (from CNRC and ONS) and existing local insurance exposure data
• National building census data used with local data for validation • Used to estimate: - National economic loss - Potential TCIP exposure and loss
• Structures are classified based on building dates - Ottoman period (Kasbah) - French Colonial period - Engineered structures post 1975 - Building codes • Used to estimate: - Potential ACIP exposure and loss - Loss to local insurance market
(Source: CAR)
12
loss estimation (1)
Estimate losses to: • National economy • Insurance portfolio - Potential - Actual
Synthetic earthquake catalogue • a set of credible earthquake events • each with an annual probability of occurrence • event based probabilistic loss estimation - results allow loss exceedance curves to be plotted
Applications: • Pricing of insurance policy • Risk control • Reinsurance Purchase • Transfer of risk
Single event: • historical events • user-defined hypothetical event • quick post-event loss estimation 13
loss estimation (2) example: ACIP and local market exposure
Risk type
Total Sum Insured
Immobilier
$$$,$$$,$$$,$$$
Commercial
$$,$$$,$$$,$$$
ACIP Total
$$$,$$$,$$$,$$$
Industrial
$$,$$$,$$$,$$$
Algerian Market Total
Percentage deductible on Loss
Minimum deductible (in DZD)
Co-insurance (propertyowner’s share)
Immobilier
2%
30,000
20%
Commercial
10%
0
50%
Industrial
10%
0
50%
Risk type
14
$$$,$$$,$$$,$$$
loss estimation (3) For illustration only Loss to ACIP 1,200,000,000
Estimated Loss (US$)
1,000,000,000
Annual Average Loss
800,000,000
600,000,000
400,000,000 Loss - ACIP Loss - residential
200,000,000
0 0
50
100
150
200
250
300
350
Return Period (Years)
Earthquake Damage Ratio 15
400
450
500
effects of an earthquake a) ground shaking b) liquefaction c) landslide d) tsunami, seiche e) fire following earthquake
16
catastrophe model coverage by main catastrophe modelling software companies
17
application of earthquake modelling • Our earthquake models are designed in conjunction with international and local academics and industry experts • In addition to hazard and vulnerability issues, we have carried out a national buildings census and provided rating recommendations • The models are used for reinsurance design and marketing activities • We have since applied earthquake modelling experience in other countries
18
The use of catastrophe models in the TCIP and other national insurance schemes Shigeko Tabuchi
6th DPRI-IIASA Forum 15 August 2006, Istanbul, Turkey
Making better reinsurance decisions…
6th DPRI-IIASA Forum 15 August 2006, Istanbul, Turkey
Making better reinsurance decisions…
about Willis • A leading global insurance broker - We are one of the world's leading risk management and insurance intermediaries - We are in the business of identifying, analyzing and managing risk - We deliver professional risk transfer, risk management, loss management and actuarial services to companies, as well as financial and employee benefits consulting.
• Company originated in 1828 • We have approximately 15,800 people around the world, with 300 offices in 80 countries • We serve both private and public clients in over 180 countries - clients include; corporations, public entities and institutions
• Willis Analytics & Solutions - The team has 150 staff in the UK, US and a network of specialists - Expertise split between: o Catastrophe Management Services o Development & Support o Financial Management Services o Marketing, Applications and Delivery o Willis Integrated Solutions 2
why model catastrophe risks ?
• Catastrophe (natural perils) risks are the major hazards that insurers are exposed to • earthquake • flood • windstorm
• The cost of natural hazards is continually increasing • Increased urbanisation • More people buy insurance
A high proportion of the population is in the highest risk zone ...and there has been significant seismic activity in the most populated areas in Turkey
3
• Every major city in the world is exposed to one or more natural hazards
why national natural catastrophe pools and programmes? • To lessen the effect of natural catastrophes on public and private finances • To achieve risk distribution and to spread the financial costs of natural catastrophe to international reinsurance and capital markets • For pools, to build up funds within a pool to reduce the requirement for external risk distribution
IDENTIFICATION
QUANTIFICATION
PRIORITISATION
PRIVATISATION
Which hazard exposures? e.g.
Impact of natural hazard
• Actual (current cost)
• Pooling mechanism
• Volcano
• Economic
• Potential cost
• Pricing
• Earthquake
• Government finances
• Recovery potential • Premium collection • Financing options • Distribution • Action plans channels
• Windstorm • Biohazard • Flood
• Social
• Hail
• Claims payment
• Drought
• Mitigation
4
EXPORT
• Structure • Price • Marketing • Leverage • Economic benefit
example of Turkish Catastrophe Insurance Pool (TCIP) • Launch:
2000
• Scope:
residential buildings
• Peril:
earthquake, flood
• Compulsory:
yes
• Cover:
originally US$ 28,000 (2000), currently approx. US$ 70,000. liable to change due to inflation and fluctuations in Turkish New Lira exchange rate
• Administration: premium collected by insurers and scheme administered by the pool manager • Risk transfer:
international reinsurance
• Guarantee:
Turkey Government (via World Bank loans)
Points of interest: • Regarded as a success, and many other countries are interested to follow its example • First project to carry out the whole process from catastrophe modelling to risk transfer for a whole country • First World Bank risk privatisation project • Take up rate for cover is average of C.15% to maximum of C.25% (in Istanbul). • Large capacity for reinsurance programme was available shortly after major earthquake event (Izmit 1999).
5
risk modelling at national level - Turkey Turkish Emergency Flood and Earthquake Recovery project (TEFER) • The World Bank wanted to avoid acting as a quasi-(catastrophe) reinsurer • The catastrophe risk modelling and establishment of an insurance pool were requirements of the World Bank, in order for emergency funds to be released after the Koecaeli earthquake in 1999 • The project was tendered according to strict World Bank rules, with a requirement for a very high “technical score” • Willis Re was appointed by the Turkish Government and the World Bank to create a new earthquake insurance model and design / reinsure a national pool
6
example of Algerian Catastrophe Insurance Programme (ACIP) • Launch:
2004
• Scope:
residential buildings, commercial buildings and contents
• Peril:
earthquake, flood, other natural catastrophe
• Compulsory:
TBD
• Cover:
under the Obligatory Insurance against Natural Catastrophe
• Administration: local insurers provide insurance cover to property owners, • Risk transfer:
Insurers have 30% (retained) to 70% (ceded) quota-share with CCR (government owned, state reinsurer) and CCR purchases reinsurance internationally
• Guarantee:
CCR does not have a reserve fund but Algerian state provides unlimited guarantee to CCR
Points of interest: • Take up rate is not as high as TCIP but slowly picking up • Covers commercial properties • High population and high seismic risk area along the northern coast • Large capacity for reinsurance programme available
7
how do catastrophe models work?
“cat model” loss estimation software
INPUT
OUTPUT
Exposure Exposuredata data
Expected Expectedlosses losses Hazard model
Inventory
Vulnerability functions
8
Financial Models
earthquake model – key inputs Key inputs: • Historical events database Source: KOERI
• Neotectonic data • Seismic source zones • Ground condition data
Active faults
Mathematical relationships:
10.00 PGA (g)
Bommer et al. 2003 Rock
• Fault rupture dimensions • Recurrence relationship - to calculate annual occurrence of an event
1.00
Bommer et al. 2003 Stiff Soil Bommer et al. 2003 Soft Soil
Seismic source zones
Bommer et al. 2003 Stiff Soil + 1s.d. Recorded PGA (geometric mean)
0.10
• Attenuation equation - to calculate ground motion or intensity values at a given location
0.01 1
10
d (km)
Attenuation equation
9
100
1000
vulnerability functions Damage / Loss experience
Engineering approach Demand Curve v Capacity Curve for building type A Spectral Acceleration Sa (g)
2003 Boumerdes earthquake
1.6 1.4 1.2 1.0
Basic Demand Curve Damped Demand Curve
0.8
Capacity Curve 0.6 0.4 0.2 0.0 0.00
Source: CRAAG
0.10
0.20
0.30
0.40
0.50
Spectral Displacement Sd (metres)
RC Lowrise, good construction MeanDamage Damage Ratio Mean Ratio
0.6 0.5
EMS C&S
0.4
CAR:RMS CAR:Willis
0.3
Radius
0.2
FH
0.1 0 5
6
7 8 MMI Intensity
9
Source: CAR
10
10
0.60
exposure information Where is it?
CRESTA Zone
- Location information (country, state/province, district, city, postcode, full address, latitude-longitude…)
What is it? - Risk type (residential, commercial, industrial) - Coverage type (Building, Contents, Business Interruption) - Building construction type, age, height,
Province
How much is it worth? - Property value - Sum Insured and Policy Count - Insurance policy conditions (deductible, limit, coinsurance)
11
District
national building stock dataset • Turkish building classified into 15 buildings types according to their known seismic performance
• Algerian building classified into 15 building types • National census data other official database (from CNRC and ONS) and existing local insurance exposure data
• National building census data used with local data for validation • Used to estimate: - National economic loss - Potential TCIP exposure and loss
• Structures are classified based on building dates - Ottoman period (Kasbah) - French Colonial period - Engineered structures post 1975 - Building codes • Used to estimate: - Potential ACIP exposure and loss - Loss to local insurance market
(Source: CAR)
12
loss estimation (1)
Estimate losses to: • National economy • Insurance portfolio - Potential - Actual
Synthetic earthquake catalogue • a set of credible earthquake events • each with an annual probability of occurrence • event based probabilistic loss estimation - results allow loss exceedance curves to be plotted
Applications: • Pricing of insurance policy • Risk control • Reinsurance Purchase • Transfer of risk
Single event: • historical events • user-defined hypothetical event • quick post-event loss estimation 13
loss estimation (2) example: ACIP and local market exposure
Risk type
Total Sum Insured
Immobilier
$$$,$$$,$$$,$$$
Commercial
$$,$$$,$$$,$$$
ACIP Total
$$$,$$$,$$$,$$$
Industrial
$$,$$$,$$$,$$$
Algerian Market Total
Percentage deductible on Loss
Minimum deductible (in DZD)
Co-insurance (propertyowner’s share)
Immobilier
2%
30,000
20%
Commercial
10%
0
50%
Industrial
10%
0
50%
Risk type
14
$$$,$$$,$$$,$$$
loss estimation (3) For illustration only Loss to ACIP 1,200,000,000
Estimated Loss (US$)
1,000,000,000
Annual Average Loss
800,000,000
600,000,000
400,000,000 Loss - ACIP Loss - residential
200,000,000
0 0
50
100
150
200
250
300
350
Return Period (Years)
Earthquake Damage Ratio 15
400
450
500
effects of an earthquake a) ground shaking b) liquefaction c) landslide d) tsunami, seiche e) fire following earthquake
16
catastrophe model coverage by main catastrophe modelling software companies
17
application of earthquake modelling • Our earthquake models are designed in conjunction with international and local academics and industry experts • In addition to hazard and vulnerability issues, we have carried out a national buildings census and provided rating recommendations • The models are used for reinsurance design and marketing activities • We have since applied earthquake modelling experience in other countries
18
The use of catastrophe models in the TCIP and other national insurance schemes Shigeko Tabuchi
6th DPRI-IIASA Forum 15 August 2006, Istanbul, Turkey
Making better reinsurance decisions…
