Understanding Britain’s Subsidence Divide

26 October 2026 / 4 min read

July 2026 was the driest month England and Wales have seen in 190 years of records. Long dry spells draw water out of shrinkable clay, the ground shrinks, and homes built on it can move. Summers like this one are becoming more common, so the trigger is arriving more often. The weather is only the trigger, though. Which homes are at risk depends on the ground, the trees and the building itself, and that was already known before the summer began. Verisk's model scores 34 million homes in Great Britain on more than forty ground, weather, vegetation and property features. It shows how much risk varies between homes that a regional or postcode average would treat as the same.

damage on concrete

Modelled risk is shown in the maps.

A London home is about five times likelier to claim than a Scottish one

London runs at roughly twice the national rate. The South East and East of England aren't far behind. Everywhere else sits at or below average, and Scotland is at less than half. Most of that pattern comes down to geology, which the next map shows.

Portfolio implication: A single national loading hides a steep concentration. The average matters less than where your addresses are.

Long-run expected claim frequency against the Great Britain average. The spread across regions is more than fivefold.

The national pattern follows the clay belt

Shrinkable clay sits under much of London, the South East and the East of England, and that is where risk is highest. Away from the clay, across most of Wales, northern England and Scotland, risk is much lower. So the ground explains the broad pattern. It doesn't explain why two homes on the same clay can have very different risk.

Underwriting implication: A hazard map shows where the clay is. It can't rank the homes built on it. The next two sections look closer to show why that ranking matters.

Lower risk Higher risk

The national view shows where risk concentrates. It is too coarse to assess individual homes.

Regional averages conceal local pockets of risk

Risk varies a great deal within London. Haringey and Barnet run at more than three times the national rate. Newham and Tower Hamlets sit at or below it. Within a single borough, neighbouring cells on this map can differ severalfold. A borough or postcode-area average combines all of that and can't tell you which homes carry the exposure.

Pricing implication: Homes in the same borough share the same weather and broadly the same geology. Their risk still differs a lot. The remaining difference comes from the property itself and its immediate surroundings.

Lower risk Higher risk

Borough outlines are for orientation only. The risk pattern does not follow them.

Property age and type change the picture again

For every house type, the oldest homes carry the most risk. But the decline with age differs by type. For detached homes it is gradual. For terraces it is steep: a mid-terrace built between 1920 and 1945 carries about a third of the risk of one built before 1837. Age explains more of the variation than type does, and the two interact, so a separate loading for each would miss part of the picture.

One comparison: A semi-detached home built between 1837 and 1870 carries more than twice the risk of one built in the 1980s.

Houses only. Warmer colours mean higher long-run claim frequency against the Great Britain average.

No single factor explains subsidence risk

The ground determines how susceptible a home is. A long dry spell is what starts the movement. Nearby trees and the building itself affect whether that movement leads to a claim. Seen this way, this summer's damage was not random. The homes that cracked already carried the risk. The weather only triggered it.

Model view: Because these factors interact, a map of the ground alone can't rank homes. Scoring every address on all four can.

model-factors

Ground, weather, surroundings and property interact. Scoring every address on all four is what gives the address-level view.

See what the model finds in your portfolio. We'll run your portfolio through the model and show where risk concentrates at address level, and where regional or postcode averages are hiding it.

Request a portfolio benchmark

Source: Verisk UK Subsidence model, build 5A (address), July 2026 release. Analysis of 34 million Great Britain residential addresses. BGS GeoSure Insurance Product used under licence.
Weather context: Gautam Naik, "Londoners Find 'Horrendous' Cracks in Their Homes After Successive Heat Waves," Bloomberg, 23 August 2026.
Risk is shown as a relative measure of long-run expected claim frequency compared with the British average. Cells containing very few homes are suppressed. Northern Ireland is outside the model's scope. These are modelled estimates, not forecasts for any individual weather year.

Britain’s Subsidence Divide

A London home is about five times likelier to claim than a Scottish one

EXPECTED CLAIM FREQUENCY BY REGION

The national pattern follows the clay belt

GREAT BRITAIN 5-KILOMETRE CELLS

Regional averages conceal local pockets of risk

GREATER LONDON - 250-METRE CELLS WITH BOROUGH BOUNDARIES

Property age and type change the picture again

RELATIVE RISK BY PROPERTY TYPE AND AGE HOUSES ONLY

No single factor explains subsidence risk