Case Study 1: Hurricane Andrew and the Carriers That One Storm Took Off the Board

A note on sourcing: this study uses the public, documented record of Hurricane Andrew (1992) and its effect on the property-insurance market. The structural facts — that Andrew was one of the costliest U.S. catastrophes of its era, that it caused multiple property insurers to fail, and that it permanently changed how the industry thinks about geographic concentration and catastrophe capital — are well-established Tier-1 history. Consistent with the book's rules, no precise loss figure, insolvency count, or carrier financial is invented here; the lessons are drawn qualitatively from what the event demonstrably showed. Where a specific carrier's internal portfolio decisions are not part of the public record, the case reasons from the pattern the storm exposed rather than from any one company's books.

Background

In late August 1992, Hurricane Andrew made landfall in South Florida as a compact but extremely intense storm, cutting a swath of near-total destruction through parts of Miami-Dade County before crossing the Gulf and striking Louisiana. In purely meteorological terms it was a relatively small storm — its zone of catastrophic wind was narrow compared to a sprawling system like Katrina thirteen years later. But it passed directly over a densely developed, heavily insured stretch of coast, and the result was one of the costliest insured catastrophes the United States had ever seen up to that point. For the property-insurance industry, Andrew was not merely a large loss. It was a revelation — the event that exposed how badly the industry had underestimated the cost of geographic concentration, and how thin the capital behind some property books truly was.

The reason Andrew matters to a chapter on portfolio management is that the insurers it destroyed were not, for the most part, destroyed by bad underwriting in the per-risk sense. Many of the homes they had written were ordinary, individually sound risks, priced at rates that looked perfectly adequate in the quiet years before the storm. What sank carriers was not the quality of any one policy. It was the accumulation — the sheer concentration of exposure in a single peril zone that one storm could, and did, strike all at once. Andrew is the canonical real-world instance of the central warning of this chapter: a book of individually sound risks can still be a single correlated bet, and concentration is a portfolio error that no amount of per-risk discipline corrects.

The insurance / underwriting issue

The portfolio failure Andrew exposed had three interlocking parts, each a direct illustration of the chapter's themes.

First, the concentration itself. Florida property insurers — especially regional and single-state carriers — had built books heavily weighted toward the same hurricane-exposed coast. Each new coastal home looked fine at the desk; summed across the book, they constituted an enormous accumulation in one peril zone (§29.3). In the language of this chapter, the carriers had diversified by count (many policies, no single one dominant) but not by geography (every policy exposed to the same storm). The law of large numbers, which they were implicitly relying on, requires independence — and a hurricane is precisely the event that makes coastal property losses move together. The diversification was an illusion, and Andrew revealed it in a single weekend.

Second, the capital and reinsurance behind the concentration were inadequate to the true probable maximum loss. Before Andrew, the industry's understanding of how bad a single Florida storm could be was, in hindsight, too optimistic — the catastrophe models of the era were primitive, and the probable maximum loss (the worst plausible single-event loss, the subject of Chapter 30) was widely underestimated. A carrier that misjudges its PML holds too little capital against it (Chapter 28) and buys too little reinsurance against it (Chapter 27). When the actual event dwarfed the assumed worst case, the capital and reinsurance that were supposed to absorb the blow were simply too small. Several insurers exhausted their reinsurance and their surplus and could not pay all their claims — they became insolvent. The promise failed, and the state's guaranty mechanism and other carriers absorbed what the failed insurers could not.

Third, the event re-priced the entire idea of writing concentrated catastrophe risk. After Andrew, reinsurers repriced Florida catastrophe cover sharply upward, the supply of capital willing to stand behind coastal concentration contracted, and the cost of holding a coastal book rose for everyone (§29.3's point that concentration is expensive every year, not just in the storm year, became impossible to ignore). The market discovered, all at once, that it had been charging too little for the catastrophe exposure embedded in geographically concentrated books — a portfolio-level pricing failure layered on top of the portfolio- level concentration failure.

What it shows

Andrew demonstrates, in the hardest possible way, several propositions this chapter argues in the abstract:

• Per-risk soundness does not imply portfolio soundness. The carriers that failed had not, by and large, written terrible individual risks. They had written too many correlated ones. This is the single most important lesson of portfolio management, and Andrew is its proof.
• Diversification by count is not diversification by correlation. A thousand coastal homes are not a thousand independent risks; they are one bet wearing a thousand costumes (§29.3). The carriers thought they had a pool; they had a concentration.
• Misjudging the probable maximum loss is a solvency error, not just a modeling one. If you do not know how bad your worst plausible single event is, you cannot hold the right capital or buy the right reinsurance against it — and the gap between assumed and actual PML is exactly where insurers fail (Chapters 27, 28, 30).
• Concentration is expensive even in the quiet years. After Andrew the market priced coastal concentration properly, and the cost of holding such a book — in reinsurance and capital — rose permanently. The lesson, painfully learned, was that a concentrated book carries a cost every year, which a disciplined portfolio manager should have been charging for all along.

Outcome

Andrew permanently changed the property-insurance industry's approach to catastrophe and concentration. In its aftermath: the modern catastrophe-modeling industry came into its own, as carriers and reinsurers recognized that they could no longer manage hurricane exposure by intuition and needed probabilistic models of the correlated loss (the subject of Chapter 30); accumulation management — explicit limits on how much exposure a book may hold in a given peril zone — became standard portfolio practice; reinsurance buying and capital adequacy for catastrophe-exposed books were re-examined across the industry; and Florida, having seen private carriers fail and withdraw, expanded public mechanisms to backstop the property market. The broad shape of these changes is well-documented Tier-1 history, even where the precise figures are beyond the scope of this study.

For the surviving and newly-formed carriers, the lesson hardened into discipline: write coastal property, yes — the social need for it is real — but manage the accumulation, know the PML, hold the capital, buy the reinsurance, and never let the book's exposure in one zone grow past what the balance sheet can survive. That discipline is exactly the portfolio appetite and the zone-cap referral of §29.7.

Lesson

The transferable lesson is the chapter's thesis, written in the most expensive possible ink: concentration risk is the portfolio failure that per-risk excellence cannot save you from, and the only defenses are measuring the accumulation, capping it, capitalizing it, and reinsuring it — before the storm sums it for you. A carrier can be right on every risk and wrong on the book. Andrew is the proof that "wrong on the book" can mean gone. It is the reason this chapter exists, the reason Chapter 30 builds the machinery to measure the very accumulation Andrew exposed, and the reason that when Harbor Steel — one more named-storm-zone account in Port Hadley — comes to the portfolio gate in §29.7, the question is not "is it a good risk?" but "does the coastal zone have room?"

Discussion questions

1. The carriers Andrew destroyed had mostly written individually sound risks. In your own words, explain how "right on every risk, wrong on the book" is possible, and why per-risk underwriting discipline is no defense against it. (§29.1, §29.3)
2. Andrew's carriers diversified by policy count but not by geography. Using the law of large numbers (Chapter 1), explain why count without independence gives no real diversification. (§29.2, §29.3)
3. Connect Andrew to capital. If a carrier misjudges its probable maximum loss, what does it get wrong about both its capital (Chapter 28) and its reinsurance (Chapter 27), and why is the gap between assumed and actual PML where insolvency lives? (§29.3)
4. After Andrew, the cost of holding a concentrated coastal book rose permanently. Explain why a concentrated book is "expensive every year, before any storm" — and what a disciplined portfolio manager should have been charging for all along. (§29.3)
5. Apply Andrew's lesson to Harbor Steel. When the account reaches the portfolio gate (§29.7), the manager asks whether the Port Hadley zone has aggregate room. Explain why that question — not the account's own quality — is the right one, and what tool (introduced in the next chapter) actually answers it. (§29.7; Chapter 30)