Housing markets under-respond to climate risk, and the households most exposed bear the cost. The organising idea is the boiling frog: a hazard rises gradually, a household forms an intent to adapt while a recent event is salient, then defers and abandons it as the salience fades. We are dreamers, not doers, so the market under-capitalises the risk. Search reorients but does not convert into moves or into capitalised prices, and the little adaptation that happens favours the well-off, so the under-response is regressive.

This programme documents that behavioural under-capitalisation across three UK-relevant perils, using the same Rightmove search and listing micro-data that let us see the intent layer realised-transaction data cannot. The three perils are parallel, not sequential, and each takes one facet of the under-response. The diagram reads each column top to bottom: the risk, the data and methods behind it, and its findings. Every box links to its detail, each findings box to the evidence behind it, and each data-and-methods box to its variable definitions, sources, and statistics.

Shared platform Rightmove property-search enquiries and listings, 2018–2025 Flood risk The intent-to-move wedge Data Rightmove search and listings Environment Agency flood outlines HM Land Registry sold prices Methods Within-town stacked event study Enquiries per active listing, difference-in-differences Wild cluster bootstrap inference Definitions and statistics Searching for safer areas after a flood is a clean null Buyer demand for affected homes dips about a sixth at months one to two, then reverts Prices do not follow, and the dip is larger from more affluent origins: a transient, regressive wedge Findings and evidence Heat waves The adaptation gap Data Rightmove listings text and attributes EPC register, matched by property HadUK-Grid temperature Methods Rolling and seasonal EPC-premium hedonic Cooling-feature text margin Grid-cell heat difference-in-differences EPC band-boundary regression discontinuity Definitions and statistics The EPC premium peaks in winter: the label rewards warmth, not summer resilience Cooling is barely advertised, with air-conditioning in a fraction of a percent of listings Chronic heat is capitalised into cooling, about two percent per degree; acute heatwave salience moves nothing Findings and evidence Coastal erosion Discounting a certain loss Data Rightmove listings and enquiries NCERM 2024 erosion-loss polygons Land Registry repeat sales and ownership Methods Proximity difference-in-differences Repeat-sale panel with property fixed effects Degree-of-capitalisation estimate Definitions and statistics A certain, dated, total loss barely prices in: the degree of capitalisation is near zero Exit is not brought forward; doomed homes list only slightly more often The uncapitalised loss lands on resident owner-occupiers, not investors: the last owner pays Findings and evidence
One platform, three perils. Each column is a paper; read it top to bottom: the climate risk, its data and methods, then its findings. Colour marks the peril. Click any data-and-methods box or findings box for its detail.