Coastal erosion is the cleanest test of whether the market prices climate risk, because it removes every excuse. The loss is certain, it is dated, and it is total: a cliff-top house has a readable remaining life and then it is gone, not merely damaged. A rational market should mark such a home down steadily toward zero as its last years approach, and backward induction says the unravelling should start early, because the final buyer will not buy, nor the one before. This paper tests that logic on the English coast, using National Coastal Erosion Risk Map projections together with full Rightmove listing histories and Land Registry sale histories.
What the price and exit data show
The certain loss barely prices in. A repeat-sales panel that follows the same property over time, so that its view, street, and character are held fixed, shows no markdown as its remaining life shortens. The estimated degree of capitalisation is statistically indistinguishable from zero, and it rejects full capitalisation by a wide margin. A doomed home with roughly fifteen years left should be shedding several percent of its value every year; in the data it is not.
Exit is not brought forward either. Homes on disappearing frontages are listed for sale only slightly more often than protected ones, sixteen or seventeen percent a year against about fourteen. Owners are not frozen, but nor is there the rush for the door that a nearing, certain loss should produce. The timing of exit, which only full listing histories can see, shows the same muted picture as the prices.
The uncapitalised loss lands on residents. The most at-risk homes turn out to be the least held by landlords, companies, or overseas owners; the last owner in the chain is typically an individual owner-occupier. Each owner appears to price only their own tenure and pass the terminal loss down the line, so it comes to rest not on a speculator but on a resident household. That is the paper’s title: the last owner pays.
These muted results are the finding, not a failure of the design. Even when ignorance cannot be blamed, because the loss is certain and dated, the market still does not price it. That is the strongest form of the programme’s thesis: under-capitalisation survives the removal of every informational excuse.
Headline estimates
| Test | Finding | Reads as |
|---|---|---|
| Repeat-sale capitalisation | No within-property markdown as remaining life falls; degree of capitalisation near zero, full capitalisation rejected | A certain loss is not priced |
| Timing of exit | Doomed homes listed about 17% a year vs 14% for protected homes | Exit deferred, not brought forward |
| End-of-chain ownership | At-risk homes least held by landlords, companies, or overseas owners | The loss falls on resident owner-occupiers |
Estimates come from a proximity difference-in-differences that separates the sea-view amenity from the erosion penalty, a repeat-transaction panel with property fixed effects, a degree-of-capitalisation estimate against the fundamental discount a certain loss implies, and an end-of-chain ownership decomposition. See the data and methods for definitions, sources, and the full descriptive statistics.