Nommez 3 avantages d’utiliser un modèle de catastrophe
- Modeled data : Insurers aren’t limited to historical claims. If an insurer has never experience hailstorm losses in Alberta, a cat model can simulate a distribution for the event, estimate potential damages, and help set a premium
- Current exposure data : Uses the latest information about population, building practices, and replacement costs to analyze frequency and severity of events
- Sensitivity testing : Parameters can be adjusted to test how sensitive results are to changes, making cat mitigation strategies more explainable to management and regulators
Nommez les éléments qui sortent généralement d’un modèle de catastrophe (model outputs)
- Event Losses : the can be broken down by region, line of business, policy or location (typically shown as event loss table and year loss table)
- Average Annual loss (AAL) (most important output)
- Loss Cost : normalize AAL by exposure, making it easier to compare risk across different portfolios
- Exceedance Probability Curves (two types : occurrence exceedance probability ; aggregate exceedance probability)
- Return period ; inverse of the exceedance probability
- Standard Deviation : measures variability in loss estimates on an event or annual basis & capture the range of possible outcomes from a single event, due to actual losses varying from average expected
- TVaR
- Intensity : qualitative description
Lorsqu’un actuaire utilise un model de catastrophe (hors de son expertise), nommez quelques éléments à valider
- Experts : Determine appropriate reliance on experts
- Understanding : Have a basic understanding of the model
- Application : Asses if the model is appropriate for the intended application
- Validation : Ensure the appropriate validation has occurred
- Use : Ensure the appropriate use of the model and its results
Nommez quelques validations qu’un actuaire qui utilise un modèle (hors de son expertise) peut effectuer
- Frequency comparison : Compare modeled event frequency to historical records and scientific consensus
- Intensity comparison : Compar modeled intensity footprints (flood depths, wind speeds) to observed historical event data
- Backtesting : Compare modeled losses to actual losses using historical in-force exposure data at the time of past events
Nommez les deux types de modèles de catastrophes présentement utilisé
- Hazard maps : Identify areas exposed to a particular peril, sowing the likelihood of an event occurring within a given timeframe. While useful for general assessment and preliminary modeling, hazard maps provide only a simplified view of risk
- Probabilistic cat modeling : Probabilistic catastrophe models go further by simulating tens of thousands of potential events, estimating how difference properties would be affected under various scenarios and translate these impact into expected financial losses.
Décrire l’approche pour la tarification du risque de tremblement de terre (EQ risk)
- Rating : Cat models develop EQ rating territories included in the EQ coverage rating algorithm. The EQ coverage is usually an endorsement that can be applied to a property policy
- Reinsurance cost : often included in property pricing for high risk earthquake regions (including as an expense)
- Additional expenses : Additional expenses associated with catastrophes and the capital needed to support them are usually included in the ratemaking process (expenses)
- AALs : used in underwriting and risk management departments to monitor concentration risk
Décrire les 3 types d’approche pour la tarification de catastrophe ‘other perils (wildfire, convective storm, flood, hail)’
- Model : Use cat model results directly for rating territories or AALs
- Blended : Combine historical experience with model outputs. Cat losses from historical data need to be trended for inflation
- Experience : For perils like wildfire and sewer back-up, many insurers rely on their own adjusted historical data due to the absence of credible industry cat models
Nommez les 3 limites de l’utilisation de l’expérience de catastrophe pour tarifier ‘other perils (wildfire, convective storm, flood, hail). - Approach Experience
- Limited data and credibility : Historical experience may be too short or sparse to reliably estimate low-frequency, high-severity events
- Lack of predictiveness : Past data doesn’t reflect future risk, especially with changing climate or exposure patterns
- Challenges in pricing and zoning : It’s hard to price new regions or define risk zone without model support
Nommez les 3 limites de l’utilisation des modèles pour tarifer ‘other perils (wildfire, convective storm, flood, hail). - Approach Model
- Model relevance and accuracy : Many models aren’t tailored to Canadian risks, making it hard to apply them reliably without adjustments
- Integration challenges : It can be difficult to consistently apply models in ratemaking and ensure they align with a company’s specific exposures
- Cost and expertise barriers : High costs and limited understanding can make model adoption and proper use challenging
Nommez les deux types d’inondation au Canada
- Coastal Flooding : Causes by storm surge from hurricanes or tropical storms, affecting areas near the coast
- Inland Flooding : Happens away from the coast. May be fluvial (riverine) or pluvial (heavy rain)
Décrire la protection offert au Canada pour les inondations. Expliquez pourquoi les assureurs vont souvent ne pas offrir la couverture au plus à risque
Many insurers offer enhanced water damage coverage as an optional endorsement. This typically include protection for two main types of risks :
- Sewer Backup : Backup for drains/pipes/fixtures tied to public sewers; sump pump or septic tank failures
- Overland Water : Fluvial, pluvial, and snow melt entering at or above group level
It might seem counterintuitive to deny flood coverage to the people who need it most, but it’s often necessary to prevent adverse selection. (the people most likely to buy it are the one most likely to claim, becomes too expensive to offer at all)
Décrire l’approche pour la tarification du risque d’inondation (Flood Risk)
- Rating : Insurers typically use the blending approach when pricing these enhanced water coverages. Historical experience + flood simulation models are used to develop rating territories/zones, which are then included in the rating algorithms and assist in underwriting
- Reinsurance cost : included in property pricing for high-risk flood regions (including as an expense)
- Additional expenses : addition expenses associated with flood coverage and the capital needed to support it are usually included in the ratemaking process
Décrire les avantages d’utiliser ‘notional exposure data’ dans le contexte de modélisation de catastrophe
Using notional data helps avoid uneven distributions or missing information that can occur in real-world data. This reduces exposure bias when calculating rating relativity factors. The general process is
1. Start with a set of Base properties : define a standard hypothetical property
2. Change One Variable
3. Run the catastrophe model : model the AAL for each version of the property.
4. Calculate relativities
