Terra insight · Market commentary

Why the Housebuilders Are Leaving London

1. “Fundamentally Broken”

Asked in October whether Taylor Wimpey would commit to further London projects, its chief executive, Jennie Daly, answered: “No. Baseline viability is, I think, fundamentally broken in London.” The reward for what she called “a very high-risk form of development” had, in her words, “just evaporated” [1].

She is not alone. Berkeley, the builder most identified with London, has cut its annual London delivery target from around 6,000 homes to 3,000, halved its workforce on London sites and, by City AM’s account, bought no new London land in more than four years [2]. The market data say the same thing. Molior counted 5,547 private homes started in London in 2025, 84% fewer than a decade earlier [3], and just 1,440 private new-build sales in the third quarter of 2026 — fewer than in the first quarter of 2009 [4].

It is tempting to read this as sentiment, or as a negotiating position with City Hall. The appraisal says otherwise.

2. The Residual Is the Whole Story

Every scheme starts with a land deal, and every land deal starts with the same sum, worked backwards. Take what the finished homes will sell for; deduct the cost of building them, the fees, the levies, the finance and the profit a board will accept. What is left is the most a developer can pay for the site — the residual land value.

The residual is a small number left over from large ones, which is why it is so sensitive. On a typical London flatted scheme the land is perhaps a tenth of the value of the finished homes. Move the costs by a tenth of the value and the land is worth nothing. When the residual goes negative there is no price at which a landowner will sell and a developer will buy, so nothing trades and nothing starts. That is what “broken viability” means.

3. One Scheme, Five Years

To see how it happened, take one illustrative scheme: 100,000 ft² of flats in outer-central London, 35% affordable, selling at £1,100/ft² and costing £350 to build on 2021 assumptions. Now change one assumption at a time to where it stands in 2026, and solve the land value each time for the profit a developer needs, measured on the value of the homes (GDV).

It is solved on two bases. A developer borrowing 65% of cost, with its own equity in first and a 1% arrangement fee, is close to how schemes are funded today. The traditional residual that valuers use finances every cost — 100% — at the debt rate, with no fee. Both borrow at 5% in 2021; by 2026 the loan costs 8% and the traditional rate is 7%. The first is solved to a 20% profit on GDV; the traditional basis, which already charges interest on every pound spent, to 15%. Each figure below is a full monthly appraisal in Terra; the four appraisals behind them are in section 7. The scheme is illustrative, not a valuation of any real site.

65% loan: equity first, 1% arrangement fee, 5% then 8%, solved to a 20% profit on GDV. Traditional: 100% finance, no fee, 5% then 7%, solved to 15%. A change may differ by £0.01m from the difference between two rounded land values.

Four things stand out. First, build cost is the biggest single blow, and it is the same blow on both bases: an 18% rise takes £6.8m off the land on a 65% loan and £6.2m on the traditional basis — the whole land value either way. Second, no other change looks fatal by itself — £1.5m here, £2.2m there — which is why each new rule could be defended as modest when it was introduced. Third, they compound. A second staircase costs saleable area on every floor; the wait for the Building Safety Regulator’s Gateway 2 approval, which London schemes have waited far longer than the twelve-week target for [5], costs interest on the money already spent and lets tender prices keep rising; and weaker sales cut the value everything else is deducted from. Fourth, the method does not change the answer. Once each basis is financed as it would be — a developer’s 65% loan with a fee, the valuer’s 100% at a lower rate — the two never differ by more than £0.4m after the build cost step, and they end 2026 within £0.1m of each other.

The second-staircase rule now applies to new residential buildings over 18 metres, with transitional protection only for schemes approved and sufficiently progressed [6]. Its cost in the table is the floor area it takes, not the redesign; a scheme that has to be redrawn after planning loses time as well.

4. What the Emergency Package Buys Back

The Government and the Mayor saw this coming. Their emergency package, adopted as London Plan guidance in March 2026, lets schemes on private land that offer at least 20% affordable housing, rather than 35%, skip the upfront viability assessment, and adds temporary relief from borough CIL [7] [8]. It runs for applications validated by 31 March 2028 and schemes started by 31 March 2030.

65% loan: equity first, 1% arrangement fee, 5% then 8%, solved to a 20% profit on GDV. Traditional: 100% finance, no fee, 5% then 7%, solved to 15%. A change may differ by £0.01m from the difference between two rounded land values.

The package wins back £7.5m on a 65% loan and £7.8m on the traditional basis, and the affordable housing cut does most of the work. On these numbers it is not enough: the land is still worth minus £1.9m on a 65% loan and minus £1.5m on the traditional basis. A scheme with higher values or a cheaper frame will cross zero, and some have — but a site bought in 2021, or valued by its owner on 2021 evidence at £6m, will not be released for a fraction of that, and planning guidance expects a landowner to receive a premium over the site’s existing use value before it is. The package narrows the gap rather than closing it, and it does so by giving up affordable homes. Nor does it touch the demand side: Daly also pointed to the overseas off-plan buyers who once let developers lay off risk before building, and who have largely gone [1]. A temporary relief does not bring them back.

5. What This Means in Practice

• For landowners, the price a site fetched in 2021 is not evidence of what it is worth now. An appraisal that starts from the land price and works forwards will show a loss; one that starts from today’s values and works back shows why.
• For developers still buying, the programme is an input, not a given. Nine months’ wait cost about £1.7m in the table above, on either basis. Model the Gateway 2 period explicitly, with its interest and its cost inflation, rather than folding it into a contingency.
• For boroughs and lenders, the order of the changes matters less than their sum. Each rule is affordable alone; it is the stack that breaks the residual. A bridge from the old appraisal to the new one, step by step, is the clearest way to show it.
• For everyone, the package is time-limited. A scheme that only works with the relief has to start by March 2030, which makes the programme part of the viability case.

6. Conclusion

The housebuilders have not lost interest in London; they have run the appraisal. When every reasonable assumption moves the same way, the residual goes negative, and a negative residual is not a hard negotiation — it is no deal at all. The useful question for anyone with a London site is not whether the market will come back, but which assumption has to move, and by how much, for the land to be worth something again. That is a question a residual appraisal answers directly, and it is worth asking before the next site is bought or sold.

7. The Assumptions, and the Appraisals

An illustrative scheme, not a valuation of any site. Not counted in the reading time above.

During the Gateway 2 wait the build cost rises with tender prices, at 4% a year. Below zero the land value is what the landowner would have to pay a developer to take the site. No stamp duty is charged on it, and the agent’s and legal fees are on Terra’s “Purchase price (cost only)”, so they are not credited back either. The 2026 steps change only the inputs named in the tables.

The appraisals

The 2021 and 2026 appraisals on each basis, as printed by Terra: the summary and the month-by-month cash flow, two pages each. The purchase price in each is the residual land value in the tables above. The full Terra models, which open in Excel with every total a live formula and reopen in Terra, are available on request from info@terra-appraisal.com.

• 2021, 65% loan — land value £5.74m
• 2026, 65% loan — land value −£9.38m
• 2021, traditional (100% finance) — land value £6.14m
• 2026, traditional (100% finance) — land value −£9.28m

On the traditional basis nobody puts equity in, so the leveraged IRR and equity multiple read “n/a”. In the 2026 appraisals the landowner pays at the start, which flatters every IRR; read the land value and the profit, not the IRRs.

8. Sources

Read 18 October 2026. Source 4 is behind a paywall; its figures are as reported in its summary.

This article is commentary for general information, not investment, valuation or financial advice. The scheme in it is illustrative; any appraisal of a real site should be prepared on current data and in accordance with the applicable professional standards.