Chapter 11. Catastrophe Modelling
Event:
Stochastic Event set with each event given a set of parameters, location and probability - may include actual historic events
Hazard:
Part of the event that causes damage - example a storm it is wind or rain
Inventory (Exposure):
Location and specific details of the insured's risks exposed to the event
Vulnerability:
Degree of loss from a specified level of hazard (damage ratios) as a percent of sum insured
Financial Analysis:
Use policy conditions to convert the losses into insured losses. Will also apply RI such that you get the following -
Ground Up
Gross of RI
Net of facultative cover, XOL and Prop (before applying Cat XOL)
Net of all RI
Aggregate Models:
Sum insured is aggregated for areas (like Crestas)
Detailed Models:
The detail of each risk exposed is fed into the model to give a more accurate outcome
Allow fo secondary uncertainty and secondary effects
Windstorm: Tropical Cyclone, Tornado and "Windstorms"
Earthquake: Shaking of the earth, liquefaction, tsunami, fire following and landslide
Path of storm, lowest pressure, size, decay and speed
Consider climate change
Consider actual trends versus better measuring equipment
Can turn demand surge and storm surge on or off
Dubiousness of wind versus flood versus storm damage
Difficult to compare to historic data and capturing systems
Key Factors:
Model iteration - tend to improve over time
Measurement over time becomes more accurate (example, number of storms increase as we are able to observe weaker storms)
Change in Demographics
Level of Building Standards
Good models follow the money (insurance saturation)
Some scales are based on effects and not the event itself, need to be aware
Quality and Quantity of historic data (events, exposure)
Difficult to model moving risks (Motor, goods in transit, marine, etc)
Amplitude, duration and frequency of ground shaking
In SA consider mining induced and natural (difficult to prove)
Different strength measurements
Very infrequent in SA, need to fill in the gaps using other countries history
Moment Magnitude, focal depth, area of fault, rate of decay
Consider sprinkler leakage, fire after and demand surge
Terrorism: Deterministic and Probabilistic
Explosives
Vehicle bombs
Aircraft hijacking
Weapons of mass destruction
Use of Catastrophe Models
Aggregate Modelling: 1 in x year return period loss estimate
Pricing: Structure and price XOL RI
Assess cat component of specific risks (both RI and Primary insurance
Allow for cat risk in exposure
Peril specific pricing for Cat
Premium loading factors to cover Cat costsAppropriate capital allocations to get correct return on capital
Capital Allocation and Assessment (SAM)
Internal capita allocation
Regulators and Rating Agencies
Used in ORSA
Reinsurance Purchase:
Vertical and horizontal cover
Reserving: Post event
Other: Cat Bonds (ILS) Pricing
This is a tool, not 00% accurate, understand the inputs and outputs
Frequency Trends: El Nino, DIPOLES, man-made effects
Severity Issues: Building codes, population trends, peril insurance take-up, change to portfolio, change to T's & C's
Secondary effects (Demand Surge)
Model assumptions and Approximations
Data issues: If Unknown, modeller uses Default, we should understand the process
Unmodelled Elements: Contracts, component of contract, unconsidered classes, elements and perils
Can gross up model, or use industry losses
Use of Different Models
Communication
Model Providers:
RMS
AIR
EQECAT