Pricing a Coast That Will Not Stay Still

AThe business of putting a price on misfortune is older than the statistics that now embarrass it. Merchants in fourteenth-century Genoa were already paying others to carry the risk of a cargo lost at sea, and the fire policies sold in London after the 1660s worked on the same principle: many contribute a little so that the unlucky few may be made whole. Two ideas hold that arithmetic together. The first is that misfortune spread across enough policyholders becomes predictable, a property known as the law of large numbers; the second, quieter and rarely stated aloud, is what actuaries call stationarity, the belief that the record of past losses remains a fair guide to the losses to come. Insured weather losses worldwide averaged about 14 billion dollars a year in the 1980s and about 71 billion in the 2010s, and no plausible account of that increase leaves the second belief standing. Whether the insurers now retreating from exposed coastlines represent a market failing or a price system working exactly as designed is what divides the field.

BIngrid Sollberg, a catastrophe-risk analyst at the Kelso Institute, has spent seven years on the Tarnholm delta, where the bills sent to the same 9,000 dwellings were followed from 2015 to 2022. Her project was commissioned to test a suspicion then common among regulators, that the flood models used by insurers overstate danger in order to justify charging more. On that question the verdict went against the suspicion: premiums rose by 41 per cent across the period, and the modelled annual loss for the delta finished within six per cent of the sums the insurers actually paid out. What the study is now cited for is something it was never designed to measure. Households facing the higher bills raised floor levels, fitted barriers and moved wiring upstairs at roughly three times the rate recorded in the comparison district, and the gap between the two districts widened rather than narrowed after the second year. Sollberg regards this as her most robust result, precisely because nobody had gone looking for it.

CThe finding carries a boundary that Sollberg states far more often than those who quote her repeat it. Every dwelling she followed was drawn from the register of mortgaged homes, on which flood cover has been a condition of lending since 2014, because that register was the only list the supervising authority allowed her to use. The study therefore says nothing about uninsured households, which are the very group on whose behalf the argument about affordability is usually conducted. One design feature does travel further than the rest: from 2016 the delta's insurers were obliged to itemise what they sent out, so that the flood charge stood on a line of its own, whereas the arrangement it took the place of had quoted one bundled figure and no more. Sollberg attributes part of the retrofitting response to that change alone, and has written twice that her figures should not be offered as evidence about what price does to homeowners in general.

DHalvard Prytz, an insurance economist at the Ferndale School of Government, does not contest the modelling result and says so plainly. His objection is that an accurate price and a workable market are different objects. Charged the premium its own risk implies, about nine per cent of the national housing stock would find no insurer willing to quote at all, against the 1.6 per cent in that position today, and no supervisor anywhere has yet let full risk-based pricing run for longer than four years. That horizon, he argues, is far too short for the decisions a permanent regime would force, such as abandoning a house outright or rebuilding a small town further inland. Put to him directly, he concedes two things: his estimate assumes that no public reinsurance stands behind the market, an assumption he agrees inflates the share, and his argument about behaviour cannot itself be tested until some government lets such a regime stand.

EGovernments have not waited for the argument to close. Since 2017 at least eleven jurisdictions have capped what insurers may charge on exposed property, and the ceilings are popular in a way the underlying arithmetic has never been. A cap holds the bill down for the household already living on the floodplain and, by the same stroke, informs every buyer who comes after it that the floodplain costs no more than the hillside. Neither Sollberg nor Prytz defends the practice. Both name it the most expensive kindness in the field, and both have argued that a legislature wishing to protect a household should send that household money directly, leaving the premium free to say what it knows. It is the rare point on which the two agree without qualification, and neither of them expects a parliament facing re-election to oblige.

FWhat is emerging is a division of labour rather than a verdict. Against perils that strike one building at a time, such as a kitchen fire, commercial cover has no substitute, because the losses of the unlucky can be met from the contributions of the many who were spared that year. Where a single storm empties a whole region's policies at once, that arithmetic collapses, and every functioning market for such perils has some public body standing behind it, a point Prytz makes and Sollberg does not dispute. The question no study has yet answered is what a risk-based premium does to house prices along a street where every owner is charged one at the same moment: the dwellings that lose cover in a study of this kind sit among neighbours whose cover continues, and prices, unlike premiums, are settled for the street rather than for the building.