Section 4 : Cats deep dive

Question 1

Why is historical loss data insufficient for catastrophe modeling?

Answer 1

Because catastrophe losses are rare and highly volatile, requiring simulations of hypothetical future events.

Question 2

!! What are the five key components of a catastrophe model?

Answer 2

1. Event Module: Defines stochastic events (e.g., hurricanes, earthquakes).
2. Hazard Module: Determines physical risk (e.g., wind speed, flood depth).
3. Vulnerability Module: Converts hazard impact into damage estimates.
4. Financial Analysis Module: Translates damages into insured losses based on policy terms.
5. Exposure Module: Represents insured properties and assets​.

Question 3

What does an Average Annual Loss (AAL) represent?

Answer 3

The expected value of catastrophe losses in a given year.

Question 4

What are Occurrence EP and Aggregate EP curves?

Answer 4

• OEP Curve: Probability that the largest event in a year exceeds a threshold.
• AEP Curve: Probability that total losses from all events in a year exceed a threshold.

Question 5

!! How do catastrophe models assist insurers in setting rates and capital requirements?

Answer 5

Pricing: Helps determine risk-based premiums by estimating probable losses.

Reinsurance decisions: Identifies optimal coverage to transfer extreme risks.

Regulatory capital: Supports solvency assessments and capital adequacy calculations.

Portfolio risk management: Assesses exposure concentration and diversification strategies​.

Question 6

What is secondary uncertainty in catastrophe modeling?

Answer 6

Variability in damage outcomes for the same event due to structural differences.

Question 7

What is Tail Value at Risk (TVAR)?

Answer 7

The average loss expected beyond a specific exceedance probability.

Question 8

How does the vulnerability module work?

Answer 8

It estimates the potential damage to buildings and infrastructure given the hazard intensity at a location.

Question 9

What is a return period in catastrophe modelling?

Answer 9

A return period estimates the average time between events of a certain magnitude.
Example: A 1-in-100-year flood has a 1% probability of occurring each year.

Question 10

How does climate change impact catastrophe models?

Answer 10

Historical data may underestimate future risks due to changing climate patterns.

New scenarios are integrated into models to adjust for rising temperatures, sea levels, and extreme weather events.

Question 11

Imagine you are an insurance company using catastrophe models. A new residential area is being developed in a flood-prone region. How would you use catastrophe models to determine the insurance premium for homes in this area?

Answer 11

I would analyze historical flood data, use the hazard module to assess flood risks, apply the vulnerability module to estimate property damage, and use financial modelling to determine expected losses and appropriate premiums.

Question 12

What is a Stochastic Event Set?

Answer 12

It is a set of simulated catastrophe events based on historical data and scientific understanding, including possible future events that may have never occurred before.

Question 13

What does the event module in a catastrophe model do?

Answer 13

It defines a database of potential disaster events, specifying their location, intensity, and probability of occurrence.

Question 14

What is the function of the hazard module in catastrophe modelling?

Answer 14

It determines the intensity of a disaster at various locations (e.g., wind speed for cyclones, ground shaking for earthquakes).

Question 15

How does the financial analysis module work in catastrophe modelling?

Answer 15

It applies insurance policy conditions (e.g., deductibles, coverage limits) to calculate the insured losses from a catastrophe event.

Question 16

Why is the exposure module important in catastrophe modelling?

Answer 16

It contains data on insured buildings and assets, helping determine how vulnerable an area is to disasters.

Question 17

!! What are some challenges faced in catastrophe modelling?

Answer 17

• Data limitations: Lack of historical flood data, especially for low-frequency events.
• Climate change impact: Increased variability and unpredictability in flood patterns.
• Model validation: Difficulty in assessing accuracy due to limited real-world comparisons.
• Regulatory constraints: Different jurisdictions have varying flood insurance requirements​.

Question 18

How do reinsurers use catastrophe models?

Answer 18

They use models to estimate potential losses, set premium prices, and determine how much reinsurance coverage to offer insurers.

Question 19

How do catastrophe models help insurers manage risk?

Answer 19

• Pricing policies based on expected losses
• Identifying high-risk areas
• Making informed reinsurance and capital allocation decisions

Question 20

Why might relying only on historical data for catastrophe modelling be problematic? Can you think of an example?

Answer 20

Climate change and urbanization alter risk factors. For example, hurricane models based on past decades might underestimate the frequency and intensity of future hurricanes due to warming ocean temperatures.

Question 21

Two catastrophe models predict vastly different expected losses for the same event. How should an insurer interpret and act on this?

Answer 21

The insurer should compare the assumptions behind the models, conduct sensitivity analysis, consult independent experts, and consider a range of possible outcomes instead of relying on a single model.

Question 22

If a catastrophe model predicts extremely high losses for a specific region, leading insurers to deny coverage or raise premiums, what ethical concerns arise?

Answer 22

This could lead to insurance inaccessibility, disproportionately impacting low-income communities. Ethical concerns include data transparency, fairness, and the potential reinforcement of socioeconomic inequalities.

Question 23

How should catastrophe models be adapted to account for climate change?

Answer 23

They should integrate updated climate projections, adjust hazard frequencies based on changing weather patterns, and consider future exposure shifts due to human development and migration.

Question 24

You work for a reinsurance company. How would you use catastrophe models to decide whether to provide coverage for a region prone to earthquakes?

Answer 24

I would assess the frequency and severity of historical earthquakes, analyze exposure data, model worst-case financial impacts, and determine whether premiums can cover risks profitably.

Question 25

!! What are the challenges in applying catastrophe models to flood insurance?

Answer 25

Data limitations: Lack of historical flood data, especially for low-frequency events. Climate change impact: Increased variability and unpredictability in flood patterns. Model validation: Difficulty in assessing accuracy due to limited real-world comparisons. Regulatory constraints: Different jurisdictions have varying flood insurance requirements​.

Question 26

What is the difference between Event Loss Tables (ELT) and Year Loss Tables (YLT) in catastrophe modelling?

Answer 26

ELT: Shows losses for each individual simulated event. YLT: Summarizes all events in a given simulated year, giving total annual losses.

Question 27

How do actuaries ensure that catastrophe models provide reliable loss estimates?

Answer 27

* Event frequency to historical occurrences. * Modelled damage distributions to real post-event data. * Intensity footprints of past disasters (e.g., wind speed, flood depth) to recorded measurements.

Question 28

Why is risk load added to catastrophe model outputs in insurance pricing?

Answer 28

Since catastrophe events have high volatility, risk load accounts for uncertainty in losses, helping insurers price policies to maintain profitability.

Question 29

How does Canada’s approach to catastrophe insurance pricing differ from the United States?

Answer 29

Canada lacks federal flood insurance; private insurers price based on historical experience + models. U.S. has government-backed flood insurance (NFIP), requiring insurers to follow state regulations.

Question 30

Why do similar buildings in the same location sometimes experience very different losses in a catastrophe event?

Answer 30

Due to secondary uncertainty, where construction quality, building age, and local factors lead to variation in damage even under identical hazard conditions.

Question 31

What is the key difference between an Exceedance Probability (EP) curve and Value at Risk (VaR)?

Answer 31

EP Curve: Shows probability of exceeding various loss thresholds. VaR: Focuses on a specific probability level (e.g., 1% VaR is the worst loss at a 99% confidence level).

Question 32

Why might traditional catastrophe models underestimate future risks due to climate change?

Answer 32

Many models rely on historical averages, which do not fully capture rising sea levels, increased storm intensity, or changing flood patterns due to climate change.

Question 33

How can catastrophe models create ethical challenges in insurance pricing?

Answer 33

* High-risk areas (e.g., flood zones) may face unaffordable premiums, leading to insurance inaccessibility. * Bias in models (e.g., outdated flood maps) can disadvantage certain communities. * Lack of transparency in model assumptions can impact policyholder trust.

Question 34

!! What are the primary outputs of catastrophe models, and why are they important?

Answer 34

1. Occurrence Exceedance Probability (OEP): Probability of the largest event exceeding a loss threshold. 2. Aggregate Exceedance Probability (AEP): Probability that total losses in a year exceed a given amount. 3. Loss Cost Estimates: Expected loss per unit of exposure. 4. Tail Value at Risk (TVaR): Measures extreme loss potential beyond a set threshold​.

Question 35

!! What factors influence the uncertainty in catastrophe model outputs?

Answer 35

* Primary uncertainty: Variability in event occurrence (e.g., where and when a flood happens). * Secondary uncertainty: Differences in loss severity given an event occurs (e.g., buildings with similar exposure experiencing different damages). * Data quality: Accuracy of input data, including property attributes and historical records. * Model assumptions: Variability in hazard frequency, intensity, and damage curves​.