Price the flood, then pay to make it smaller

I believe a European property insurer has three honest options for flood risk and one dishonest one. It can price the hazard openly and help customers reduce it. It can decline the risk and say so. Or it can contribute to a public arrangement for losses beyond what private capital can sensibly carry. The dishonest option is the one most of our industry practised for decades: a flat loading that charges the riverside business and the hilltop business almost the same, with the difference quietly paid by customers who never knew they were paying it.

That practice is coming to an end, and it is ending badly. After large flood events, insurers tighten terms, raise deductibles and in some cases stop renewing cover in the most exposed areas. Each decision may be defensible on its own. Together they widen what EIOPA calls the protection gap: the share of catastrophe losses that nobody has insured. The gap does not disappear when insurers leave. It moves to households, small businesses and public budgets, usually without anyone having decided that it should.

Silence is the most expensive pricing decision

A flat flood loading looks like solidarity. It is closer to an undisclosed transfer. When a tariff does not distinguish between hazard zones, low-risk customers subsidise high-risk ones, and nobody involved can see the size or direction of the subsidy. The customer in the riverside building receives no signal that their location carries a cost, and therefore no reason to invest in protection. The customer on higher ground pays for a risk they do not run, and eventually leaves for an insurer that prices more sharply. The book drifts towards the exposed risks, and the flat loading becomes inadequate precisely where it matters most.

I saw this from the reinsurance side for almost six years. The insurers who priced flood by zone knew their accumulations and could explain their losses. The ones who did not were often surprised by their own exposure after an event. I came to think of an unpriced hazard not as a kindness to customers but as a gap in the insurer's knowledge of its own book.

At Talvermark we changed this for small-business property after the July 2021 floods. We re-rated 41,000 policies in high-hazard zones over two renewal cycles. The average increase in the highest band was 34%. That is a painful number, and I do not pretend otherwise. But every affected customer and their broker received the zone, the loading and the criteria for reducing it. A price that can be explained can also be challenged, corrected and, crucially, reduced by the customer's own action.

Retention in the highest band fell from 89% to 81%. Some of those businesses found cheaper cover elsewhere, often from insurers who had not yet re-rated, and some of them will be uninsured or underinsured at the next event. I take no satisfaction in that. But the alternative was to keep underpricing knowingly, and an insurer that does so is making a promise it has not funded.

A prevention credit is a claim paid in advance

Pricing alone is a poor answer if it offers the customer nothing except a larger bill. The second part of our approach was a prevention credit of up to 15% for verified measures: demountable flood barriers, plant and electrics raised above expected water levels, fit-outs that can be dried rather than replaced. We spent €9 million funding surveys so that small businesses did not have to pay for an engineer before they knew whether the credit was worth having.

I think of the credit as part of the claim we would otherwise pay later. The economics are straightforward if the measures work. In a 2024 flood affecting our Austrian portfolio, the average claim at protected sites in a sample of 212 claims was 58% lower than at comparable unprotected sites. The sample is small and one event proves little, so I treat it as encouraging rather than conclusive. It nonetheless matches what engineers have long said about where flood losses come from: water reaching equipment, stock and finishes that could have been moved or protected.

Prevention has a structural problem that insurers cannot solve alone. The party who can install protection is often not the party who buys insurance. A tenant pays for contents and business interruption cover but cannot alter the building without the owner's consent. The owner insures the structure but may not care about the tenant's stock. We learnt this late. A meaningful share of the businesses that asked about the credit could not use it. I now think insurers should work with landlords, chambers of commerce and municipalities to design measures that cross that boundary, and should price owners' and tenants' cover in ways that reward cooperation between them.

There is also a limit to what individual measures can achieve. A barrier protects a doorway, not a district. Where flood defences are inadequate, building-level prevention reduces losses at the margin while the underlying hazard remains. That brings me to the role of the state.

Where the state belongs in the stack

Several European countries already combine private insurance with public arrangements for natural catastrophes. France attaches natural catastrophe cover to property policies under a regime with state-backed reinsurance. Germany has debated for years whether natural hazard cover should become compulsory for homeowners. At European level, EIOPA has published options that include a public-private reinsurance layer. The debate is no longer whether the state has a role, but where in the layers of risk that role should begin.

My answer is that the state belongs at the top of the stack, not the bottom. Private insurers should carry the frequent and moderately severe losses, priced by hazard, because that is where price signals change behaviour. Public arrangements should absorb the extreme tail, the losses so large and so correlated that private capital would have to hold uneconomic amounts of it or leave the market. A public layer placed too low removes the price signal and invites building in places that should not be built on. A layer placed too high leaves insurers withdrawing from exactly the regions that most need cover.

Compulsory cover deserves a careful answer. It would widen the pool and reduce adverse selection, and I support it for homeowners if it is paired with hazard-reflective pricing and a credible public backstop. Without hazard pricing, compulsion simply makes the undisclosed subsidy universal. Without a backstop, it obliges insurers to write risks that their capital cannot carry.

None of this lets insurers off the hook. We have to price honestly, share our hazard data with public authorities planning defences, and stop treating withdrawal as a pricing tool we never have to explain. I would ask any insurance board to look at its highest flood zone and answer three questions. What are those customers paying, compared with the modelled cost of their risk? What can they do to reduce it, and who pays for that work? If we stopped renewing them tomorrow, who would carry their next loss? An insurer that cannot answer all three is not managing flood risk. It is waiting for the next one.