Catastrophe Bond

A catastrophe bond turns insurance into an investment. An insurer facing, say, Florida hurricane risk issues a bond; investors' money sits in a collateral account earning money-market rates plus a fat risk spread (recently 5–15%). If the defined disaster happens — a hurricane causing more than $X of losses — investors' principal pays the insurer's claims. No disaster: full principal back after 3–4 years.

The magic word is uncorrelated: hurricanes don't read the Fed's minutes. A cat bond's payoff depends on wind speeds and earthquake magnitudes, not recessions — making ILS one of the very few return streams truly independent of every other page in this atlas (2022 proved it: stocks and bonds both crashed; cat bonds returned ~−2%, then +20% in 2023, their record year).

The catch is honest and simple: occasionally the disaster happens, and you lose real principal at the worst humanitarian moment. You are the reinsurer now.

Structure

• SPV issuance: proceeds sit in collateralised trust (T-bill/money-market) — no issuer credit risk, pure event risk.
• Triggers (the crucial fine print): indemnity (issuer's actual losses — sponsor-friendly, slow), industry loss (index like PCS), parametric (physical measurements: wind speed at stations, quake magnitude — fast, transparent, but "basis risk" if your losses differ from the trigger).
• Attachment/exhaustion: principal erodes between loss thresholds, like a CDO tranche on nature.
• Extension periods: after qualifying events, maturity extends while losses develop.

The market

~$45–50bn outstanding, part of the broader ~$110bn ILS/collateralised-reinsurance space; peak issuance after big hurricane years, when reinsurance prices ("hard market") spike. Spreads quote as multiple of expected loss: EL 2%, spread 8% → 4x multiple — the market's price of catastrophe risk aversion.

What investors watch

Seasonality (hurricane season prices in live), model updates (RMS/Verisk revisions move marks), climate trend loading, and "loss creep" (Hurricane Irma's losses grew for two years). Secondary trading is thin but real; dedicated ILS funds (Fermat, Twelve, Schroders ILS…) intermediate for institutions.

Pricing: actuarial science meets risk premia

Expected loss comes from vendor catastrophe models — stochastic event sets (10k+ simulated hurricane seasons), vulnerability curves, exposure databases:

The multiple \(m\) (historically 2–5x) is the tradable object: it spikes after major events (capital destroyed, fear high — the best vintages) and compresses in calm inflows. Peak-peril US wind pays the highest multiples; diversifying perils (Japan quake, Euro wind) price tighter for portfolio reasons — the market has its own CAPM where "beta" is Florida.

Model risk is the risk

All pricing keys off models whose tail calibration is unprovable: climate non-stationarity, exposure growth in coastal zones, demand surge post-event, and secondary perils (wildfire, severe convective storm) that models historically underweighted — the 2017–2022 "surprise" loss years came largely from modelled-lightly perils. Sophisticated buyers run their own view-of-risk adjustments on vendor EL before bidding.

Portfolio mathematics

Cat risk is jump risk: returns = steady carry minus rare large losses — negatively skewed but genuinely zero-beta. Allocation sizing uses tail-contribution measures (TVaR at 1-in-100 aggregate) rather than volatility; correlation to markets appears only through post-event reinsurance-price channels and money-market collateral returns. The 5–10% institutional sleeve exists because nothing else in this atlas offers payoffs drawn from the atmosphere instead of the economy.