Increasing property policies by 50% in a fast-growing coastal area
Insurance
These aren't separate policies: it's one risk, that a single hurricane hits all at once.
In the scenario Avoided a concentration that a single hurricane could have turned into a capital loss.
The protagonist & the context
The head of property catastrophe underwriting, with a 20% premium growth target and a coastal market in strong real-estate development.
A mid-size insurance company with a diversified national property portfolio. The coastal area in question has seen residential property prices grew 45% over the past five years, driven by new tourist and residential developments. The in-house catastrophe models are updated annually but calibrated on thirty-year historical series.
Background
The intensification of extreme weather events over the past decade has made catastrophe models calibrated on pre-2010 history systematically optimistic: global insured losses from climate-related events rose 180% in real terms between 2000 and 2023. Low-lying coastal areas are particularly exposed to an amplification effect: a single hurricane season can hit the entire territory within a few kilometres with one event, eliminating the geographic diversification benefit that portfolio models assume. On the reinsurance side, several global reinsurers have already reduced available capacity for high-climate-risk coastal areas, with premium increases of 30–60% in the last three renewal rounds.
The dilemma
Premiums in the coastal area are among the highest in the portfolio; expected losses — per the internal models — are within capital limits. Everything looks fine. The problem is that the models look at the past, and the past is no longer the present: with events 30% more intense than historical records, losses in an extreme scenario exceed twice what the models project. And there is a second, less visible problem: concentrating the +50% in the same coastal strip means a single hurricane hits the entire portfolio increment at once. This is not single-policy risk — it is portfolio risk that per-policy models do not capture. The reinsurance that covers this excess may not be available, or may cost too much, exactly in the season when it is needed.
Exhibits
Probable maximum loss estimate in the 1-in-100-year scenario using the historically calibrated model and with the climate adjustment applied (illustrative data).
Share of the zonal portfolio concentrated within five kilometres of the coastline — nearly fully correlated in a hurricane scenario (illustrative data).
The contradictor's analysis
- Catastrophe models capture the real risk, climate included.
- Reinsurance will remain available and at the same price.
- The policies are weakly correlated with each other.
Catastrophe models look at the past, but the climate is no longer the same: extreme events are more frequent. Concentrating the +50% in a single coastal area means putting all your eggs in one basket: a single hurricane hits the entire portfolio at once. And reinsurance — your safety net — can withdraw or get more expensive just when you need it.
- With events 30% more intense than the historical record, how much do you lose in extreme scenarios (1 in 100 and 1 in 250 years)?
- How much exposure is concentrated within a few kilometers of the coast?
- What happens if reinsurance disappears or doubles in price?
- How much do you lose, net, if a single event hits the whole area at once?
- How much of the plan depends on reinsurance that might not be there?
Provenance: climate-adjusted catastrophe models · geographically mapped exposure · reinsurance terms · red-team base.
Resolution & consequences
Methodological note
Composite teaching case. Figures and details are illustrative; it is not a client outcome or a performance claim. Built, in the method of the Harvard Business Review, from recurring decision patterns in the sector, merged and anonymized to protect confidentiality. The «provenance» notes describe the type of evidence the engine cites with traceability in production. No Δ-CSI score is shown: composite cases have no frozen, reproducible scorecard.