Chapter 30 — Key Takeaways
A one-page field card for Catastrophe Modeling and Accumulation Management. All figures illustrative.
The core claims
- Catastrophe breaks the law of large numbers. It destroys independence (Ch. 1): one cause hits a whole region at once, so a book that looks diversified by count is one correlated bet. The familiar tools (frequency × severity, the average year) describe the body of the distribution; catastrophe lives in the heavy tail.
- The cat model simulates the tail you can't observe. Tens of thousands of plausible events run through three modules: hazard (where/how intense — the peril's footprint), vulnerability (damage functions = COPE turned into math), financial (your deductibles, limits, and reinsurance turning damage into net loss).
- The EP curve holds it all. From one exceedance-probability curve you read both governing numbers — and you never confuse them.
- Read the return period correctly. "1-in-100-year" = 1% chance every year, not a once-a-century schedule. Two can strike in a decade. Over a 30-year mortgage: ~26% chance of at least one.
- Accumulation is geographic. Manage exposure by peril zone; judge each account on its marginal contribution to the zone PML, not its standalone quality.
- The baseline is moving. A backward-looking model understates risk for perils climate is intensifying — a known-sign error. Treat the historical number as a floor, not a center.
- The protection gap is the edge of insurability. Insurability is a property of the risk plus the available machinery (models, pricing freedom, reinsurance, mitigation, public backstops) — not of the risk alone.
AAL vs. PML — never confuse them
| AAL (average annual loss) | PML (probable maximum loss) | |
|---|---|---|
| What it is | the mean of the whole EP curve | a loss far out in the tail, at a chosen return period |
| Question | "What do we charge?" | "What must we survive?" |
| Lives in | the rate (the cat pure premium) | the balance sheet (reinsurance + capital) |
| Get it wrong → | underprice the average → slow death | underestimate the tail → sudden death |
The rule of thumb
Price to the AAL; capitalize to the PML; write to the zone. An account can have its AAL fully loaded into an adequate price and still be declined because its marginal PML contribution breaks the zone limit. Defend the right thing: AAL answers a pricing objection, PML answers an accumulation objection.
Return-period arithmetic
Chance of at least one 1-in-100-yr event over N years = $1-(0.99)^N$ → 10 yrs ≈ 9.6% · 30 yrs ≈ 26% · 50 yrs ≈ 40% · 100 yrs ≈ 63% (never 100%).
Key terms
catastrophe model · probable maximum loss (PML) · average annual loss (AAL) · return period / exceedance probability · accumulation management · the protection gap
What you could defend to your manager
"Harbor Steel passed the catastrophe screen. Its AAL contribution confirms the cat load in the price is adequate, and its marginal PML contribution fits within the Port Hadley zone aggregate net of the cat XOL treaty — the 5% named-storm deductible and the cession keep its net contribution modest. But headroom in that zone is limited, and because the model is backward-looking against a warming climate, I'm treating the historical PML as a floor and flagging the climate trend and the zone tightness for the next renewal."