in simple words about catastrophe modelling€¦ · use of the catastrophe models catastrophe...
TRANSCRIPT
0Aon Benfield | Proprietary & ConfidentialRisk. Reinsurance. Human Resources.
Aon Benfield
In Simple Words AboutCatastrophe Modelling21 June 2018
1Aon Benfield | Proprietary & ConfidentialClient Name, Proposal # | Practice Group | Date (##.##.##) | Document # OptionalProprietary & ConfidentialAon Benfield | Proprietary & Confidential
1. Catastrophe Perils
2Aon Benfield | Proprietary & Confidential
Natural Catastrophes
“A natural catastrophe is a major adverse event resulting from natural processes of the Earth. It can cause loss oflife or property damage. Severity of the damage depends on the affected population’s and property’s resilience,or ability to recover and also on the infrastructure available.“ (source: Wikipedia)
Catastrophe Perils
WildfiresHydrological§ Floods§ Tsunamis
Meteorogical§ Thunderstorms§ Hailstorms§ Tornadoes§ Cyclonic storms§ Droughts§ Heat waves
Geological§ Earthquakes§ Volcanic
eruptions§ Landslides
Space disasters§ Solar flair§ Asteroids
3Aon Benfield | Proprietary & Confidential
WindstormCatastrophe Perils
§ Typical features:– Large territories affected– Low damage ratios, serious structural damage is rare– Low number of casualties– Multi-country losses in Europe
§ Main windstorm territories: Western European winterstorms, US hurricanes.
§ US Hurricane season: June – October§ European Windstorm season: October – March
4Aon Benfield | Proprietary & Confidential
EarthquakeCatastrophe Perils
GSHAP seismic hazard map
Kobe, Japan 1995Izmit, Turkey, 1999
§ Typical features:– Smaller territories affected– Usually only single country losses– Damaging earthquakes are less frequent than floods
and windstorms. Typical – low frequency and highseverity
– Serious structural damage can occur. Usually highnumber of casualties
§ Main earthquake territories (RI point of view): Japan,California
5Aon Benfield | Proprietary & Confidential
FloodCatastrophe Perils
Source: Aon Benfield Impact Forecasting
§ Typical features:– Flood propagates along river streams and cannot affect
large areas continuously– Serious structural damage is not common– Low number of casualties– Typical – multi-country losses in Europe– Loss prevention can be very effective – flood defences,
early warning§ Main flood territories (RI point of view): Western Europe,
CEE, UK
Modelled flood extent
6Aon Benfield | Proprietary & Confidential
Global Insured Catastrophe Losses
Global Insured Losses – All natural disasters
Catastrophe Perils
Insured and non-insured losses
Source: Aon Benfield Impact Forecasting Source: Aon Benfield Impact Forecasting
§ 2017 was an exceptional year in terms of catastropheactivity– Overall amount of insured losses: USD 134 bn
§ Average 2000-2017: USD 51 bn§ Available capital of global reinsurers: USD 600 bn§ Annual proportion of non-insured losses: 60-80%
7Aon Benfield | Proprietary & Confidential
Hazard and Risk
Shenzen, China
Catastrophe Perils
MiamiBeach 1926
8Aon Benfield | Proprietary & ConfidentialClient Name, Proposal # | Practice Group | Date (##.##.##) | Document # OptionalProprietary & ConfidentialAon Benfield | Proprietary & Confidential
2. Catastrophe Modelling –History and Applications
9Aon Benfield | Proprietary & Confidential
History of the Catastrophe ModellingCatastrophe Modelling – History and Applications
Catastrophe model 200 years ago§ First attempts to control the catastrophe risk areas old as the property fire insurance
§ In early 19th century insurers were using tacks ona map to control the accumulations of insuredrisks, that are critical in case of a fire event
§ The mapping approach was used until 1960’s
§ In the late 1980’s the first commercial modelvendors entered the market in US
10Aon Benfield | Proprietary & Confidential
§ Perils covered by the catastrophe models: Earthquake, Flood, Windstorm, Hail, Wildfire, Terrorism
Catastrophe Modelling Software TimelineCatastrophe Modelling – History and Applications
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016
Cartograph
RiskLink
AIR – CATRADER / CLASIC/2
EQECAT – Worldcat Clasic CoreLogic
Oasis
19881986
Touchstone
RMS - IRAS ONE
KC & Co.
Aon Benfield Impact Forecasting
JBA
Worldcat Enterprise RQE
2018
§ Commercially available Cat models appeared in late 1980’s in USA§ Significantly increased demand for Cat modelling after 1992 hurricane Andrew when the US market saw
11 insolvencies§ Katrina 2005 – only 3 insolvencies despite considerably higher financial impact of the event
Regulatory Requirements
11Aon Benfield | Proprietary & Confidential
Impact Forecasting – Cat Models Development in AonCatastrophe Modelling – History and Applications
§ Impact Forecasting team based in 5 Aon offices: Chicago, London, Prague, Singapore, Bangalore§ Different groups of experts working on the models development:
– Hazard Development – experts with meteorological, hydrological and geophysical background, special team dedicated tothe terrorism modelling
– Vulnerability development – experts with engineering background dealing with the response of insured risks to the naturalperils
– Software development and support – IF modelling software and additional products are currently licensed to ~60 licensees
12Aon Benfield | Proprietary & Confidential
Use of the Catastrophe ModelsCatastrophe Modelling – History and Applications
Loss
RP (years)
1in250 Cat capacity
Probability vs modelledloss → the most common
output of thecatastrophe modelling
Catastrophe treaty can behardly structured and
placed without themodelling
§ Reinsurance was the first segment of the insurance industry that adoptedthe Catastrophe models.
§ Typical reinsurance related questions:– Is my portfolio exposed to natural hazards?– What is the main peril and what are the losses I can expect?– What protection (limit) do I need?– What price should I expect?
13Aon Benfield | Proprietary & Confidential
Can we Use the Company Loss Experience instead of the Model?Catastrophe Modelling – History and Applications
1in50 PML = 13 mil
1in50 PML = 100 mil1in50 PML = 35 mil
20 years of loss experience§ Extrapolation of historical lossdata is one of the possibilities ofhow to design the reinsurancestructure without a model
§ The disadvantages of loss dataapproach:– The data is not available for
sufficiently long period– The past losses need to be
properly indexed– Different ways of extrapolation
can lead to substantiallydifferent results
§ All the issues above are bypassedwhen using the Catastrophemodel
14Aon Benfield | Proprietary & Confidential
Advantages of a Catastrophe Model
§ The catastrophe model takes into account a wide range ofscenarios from the small events occurring on an annual basis tothe extremes without any analogues in the historical observations
§ The modelling is always done for the actual exposure, it thereforereflects the actual position of the insurance company and there’sno need for indexation of the past losses
§ The model can further take into account:– Changes in the regional distribution of the exposure– Changes in the portfolio structure per LOB (householders,
industry, CAR/EAR, BI…)– Model properly takes into account the policy conditions –
deductibles, sublimits…..
§ Models can often reflect multiple lines of business: property, BI,workers compensations, marine cargo, life, cyber
§ With the growing computational power, the modern catastrophemodels can be used in many other areas beyond the traditionalneeds of the reinsurance
Catastrophe Modelling – History and Applications
15Aon Benfield | Proprietary & Confidential
Examples of Reinsurance Related ApplicationsCatastrophe Modelling – History and Applications
§ The total sum insured of my portfoliowent up by 15%, however, it remainsstable in the Absheron region, whereI already have 50% of the exposure.Do I need 15% higher Cat limit?
§ Expected model result: The 1in250PML which drives the capacity willremain stable, however, we’ll observegrowth of losses at low return periods(e.g. <1in50 years) due to thegrowing diversification
Example 1
§ My portfolio remains stable since lastyear, however, I lost somehouseholders business, which wascompensated by the new successfulproduct covering BI. Do I need tochange the Cat limit?
§ Expected model result: very likelythere will be need for the higher Catlimit, as the BI exposure cangenerate substantially higher lossescompared to property damage
Example 2
§ The total sum insured of my portfoliowent up by 20%, however, we newlyintroduced the deductible whichcorresponds to 1% of sum insured.What will be the implication forthe reinsurance treaty?
§ Expected model result: the requiredgrowth of the Cat limit will beconsiderably lower than 20% due tothe deductibles
Example 3
16Aon Benfield | Proprietary & Confidential
Catastrophe Model as an Underwriting ToolCatastrophe Modelling – History and Applications
§ Probabilistic models based on high resolutionhazard maps can be used for location-wiseNat Cat loss assessment
§ The model can indicate how much premiumshould the insurer charge in a given locationfor a given risk type
§ Differentiated by property parameters such as:– Occupancies– Constructions– Number of stories classes– Presence of basement
17Aon Benfield | Proprietary & Confidential
Catastrophe Model and Insurance Product DesignCatastrophe Modelling – History and Applications
Effect of limits for a given postal code or Lat&Lon square
Effect of deductibles for a given postal code or Lat&Lon square§ Models can be used to determineeffect of insurance conditions– Deductibles– Limits
§ Key for attractive & profitableproduct design– Full limit possible in low risk areas– Effective limits enable insurance
within high risk zones
18Aon Benfield | Proprietary & ConfidentialClient Name, Proposal # | Practice Group | Date (##.##.##) | Document # OptionalProprietary & ConfidentialAon Benfield | Proprietary & Confidential
3. How the Cat ModelWorks
19Aon Benfield | Proprietary & Confidential
Damage
Calculation
VULNERABILITY
HAZARDEvent set
Physically modelled extents
LOSS
Insured loss calculation
Policy Conditions
EXPOSURE
Risk Characterisation
Risk Location
Catastrophe Model Anatomy
Hazard Component
How the Cat Model Works
Historical information about the past events, theirfrequency and severity, is a basis of the modelhazard component
Vulnerability Component
Damage functions – convert hazard severity(earthquake intensity, wind speed, etc.) into thedamage
Exposure Component
Information about the position of an insured risk
Loss Component
Derives loss exceedance probability curves andapplies the policy conditions
20Aon Benfield | Proprietary & Confidential
5. Repeat (for every location andevery simulated event), sum upby event to get total loss
Catastrophe Model – How it Works?How the Cat Model Works
Event ID Total loss
1 2,200,000
2 3,400,000
3 5,000,000
… …
120,000 1,200,300
4. Loss Calculation
Ground Up Loss1,000,000 * 0.5 = 500,000 AZN
Gross Loss500,000 – 20,000 = 480,000 AZN
3. Link with damage(vulnerability) function
Intensity
Dam
age
Rat
io
MSK 9
50%
6. Calculate ExceedanceProbability curve
Years
Loss
200
10bn
1. Insured location #1
Location: 40.0°N 49.0°ETIV = 1,000,000 AZNDeductible = 20,000 AZN
2. Link with EQ intensityfootprint
MSK Intensity = 9
21Aon Benfield | Proprietary & Confidential
Historical Hazard Information
Historical earthquakes in Azerbaijan – country withhigh seismic activity
How the Cat Model Works
Historical earthquakes in Hungary (country with lowseismic activity)
§ Historical information is always a basis of a stochastic event set§ Uncertainty is increasing with decreasing hazard activity. In other words it is much easier to built an
earthquake model in high activity country.
22Aon Benfield | Proprietary & Confidential
§ Earthquake intensity is converted into thedamage based on the damage functions
§ Different risk types behaves differently in thecatastrophe models
Damage CalculationsHow the Cat Model Works
Intensity
Dam
age
ratio
Industrial Agricultural
ResidentialResidential
23Aon Benfield | Proprietary & Confidential
Working with Catastrophe Exposure Data
Economic region (population density map)
How the Cat Model Works
District (Rayon)
§ Catastrophe models very often work withsimplified exposure data
§ The model must be ready for different level ofdetails to be able to produce plausible results evenfor simplified data
24Aon Benfield | Proprietary & ConfidentialClient Name, Proposal # | Practice Group | Date (##.##.##) | Document # OptionalProprietary & ConfidentialAon Benfield | Proprietary & Confidential
4. Conclusion
25Aon Benfield | Proprietary & Confidential
Modelling of Catastrophe Losses
§ Goal of Cat Modelling is to estimate possiblelosses while minimising uncertainties
§ Similarity to crossing the street– When you want to cross street (Reality) you
don’t have to check for traffic = you can justtake the risk
– Using traffic light (Model) help you to minimiserisk
– But you still think and use all available info/tools before you cross the street, i.e. you don’ttake the traffic light as the only true source ofinformation
Conclusion
26Aon Benfield | Proprietary & Confidential
Disclaimer
Legal Disclaimer©Aon Limited trading as Aon Benfield (for itself and on behalf of each subsidiary company of Aon Plc) (“Aon Benfield”) reserves all rights to thecontent of this report (“Report”). This Report is for distribution to Aon Benfield and the organisation to which it was originally delivered only.Copies may be made by that organisation for its own internal purposes but this Report may not be distributed in whole or in part to any thirdparty without both (i) the prior written consent of Aon Benfield. and (ii) the third party having first signed a “recipient of report” letter in a formacceptable to Aon Benfield. Aon Benfield cannot accept any liability to any third party to whom this Report is disclosed, whether disclosed incompliance with the preceding sentence of otherwise.To the extent this Report expresses any recommendation or assessment on any aspect of risk, the recipient acknowledges that any suchrecommendation or assessment is an expression of Aon Benfields’ opinion only, and is not a statement of fact. Any decision to rely on any suchrecommendation or assessment of risk is entirely the responsibility of the recipient. Aon Benfield will not in any event be responsible for anylosses that may be incurred by any party as a result of any reliance placed on any such opinion. The recipient acknowledges that this Reportdoes not replace the need for the recipient to undertake its own assessment.The recipient acknowledges that in preparing this Report Aon Benfield may have based analysis on data provided by the recipient and/or fromthird party sources. This data may have been subjected to mathematical and/or empirical analysis and modelling. Aon Benfield has not verified,and accepts no responsibility for, the accuracy or completeness of any such data. In addition, the recipient acknowledges that any form ofmathematical and/or empirical analysis and modelling (including that used in the preparation of this Report) may produce results which differfrom actual events or losses.The Aon Benfield analysis has been undertaken from the perspective of a reinsurance broker. Consequently this Report does not constitute anopinion of reserving levels or accounting treatment. This Report does not constitute any form of legal, accounting, taxation, regulatory oractuarial advice.