The number of old loans keeps rising, even as newer loans repay faster.
The share of specialist development loans past two years has about doubled since 2019. Some of that comes from unfiled paperwork and fewer new loans, but the count of overrunning loans is rising too.
- Level. 2,968 of 4,625 live specialist charges were past 24 months at 30 September 2026.
- Control. The wider register moved from 79.9% to 76.5% over the same period, so this ageing is specific to development lending.
- Flow. The 2024 cohort is repaying faster than 2023, which puts the difficulty with loans written earlier.
These figures can only show that a charge is still registered. They cannot show that a loan is in default, and many will have been extended by agreement. We read them as a measure of how many developers are running beyond the plan they first financed.
64.2% of live specialist development loans are over two years old
Development loans are usually written for 18 to 24 months. At 30 September 2026, 2,968 of the 4,625 live charges held by specialist development lenders had passed 24 months, against 31.9% at the end of 2019.
Companies House charge register, Construction Capital analysis
View the chart data
| Quarter-end | Specialist % | Development-like % | Whole register % |
|---|---|---|---|
| 2019-12-31 | 31.9 | 43.2 | 79.9 |
| 2020-03-31 | 34.6 | 44.6 | 79.9 |
| 2020-06-30 | 38.5 | 47.7 | 80.4 |
| 2020-09-30 | 43.4 | 51 | 80.7 |
| 2020-12-31 | 45.3 | 53.2 | 81 |
| 2021-03-31 | 46.1 | 54.1 | 80.9 |
| 2021-06-30 | 46.4 | 55.2 | 80.6 |
| 2021-09-30 | 46.7 | 56.8 | 80.6 |
| 2021-12-31 | 48.4 | 58 | 80.6 |
| 2022-03-31 | 47.9 | 58.3 | 80.4 |
| 2022-06-30 | 45.4 | 56.8 | 79.6 |
| 2022-09-30 | 43.1 | 55.6 | 79 |
| 2022-12-31 | 44.4 | 56.3 | 78.6 |
| 2023-03-31 | 47.6 | 57.6 | 78.8 |
| 2023-06-30 | 49.8 | 59.3 | 79.2 |
| 2023-09-30 | 52.8 | 60.3 | 79.5 |
| 2023-12-31 | 54.9 | 61.5 | 79.8 |
| 2024-03-31 | 56.9 | 61.9 | 80 |
| 2024-06-30 | 58.4 | 62.2 | 80.2 |
| 2024-09-30 | 61.1 | 62.7 | 80.3 |
| 2024-12-31 | 61.4 | 62.4 | 80.1 |
| 2025-03-31 | 63 | 62.1 | 79.7 |
| 2025-06-30 | 63.3 | 61.7 | 79.3 |
| 2025-09-30 | 63.7 | 61.5 | 78.7 |
| 2025-12-31 | 64.6 | 61.1 | 78 |
| 2026-03-31 | 63.8 | 60.8 | 77.3 |
| 2026-06-30 | 63.9 | 61 | 76.8 |
| 2026-09-30 | 64.2 | 61.4 | 76.5 |
The control is what makes this a development story. The whole Companies House register is older than the development book, but it has not aged: 79.9% of live charges were over 24 months old at the end of 2019 and 76.5% at 30 September 2026. Long-dated term debt is meant to sit there. A development loan is not. The wider development-like book, which adds other development-active lenders where a charge shows signs of development, has moved from 43.2% to 61.4%. Restrict that book to borrowers with a housebuilding code and the move is from 43.2% to 65.4%, so the finding does not depend on the definition.
The share has levelled off rather than turned. 64.2% is just below the high of 64.6% at 31 December 2025, but the count of 2,968 is the highest in our series.
Two forces are behind the specialist figure, and only one of them is loans overrunning. New origination has slowed, from 1,230 charges in the 2022 cohort to 938 in 2024, and fewer young loans mechanically lift the share of old ones. But the count has grown as well as the share: live charges past their lender's typical term rose from 2,779 to 2,968 in a year, and from 742 in 2019.
Companies House charge register, Construction Capital analysis
View the chart data
| Quarter-end | Live charges | 12 months or under % | 12 to 18 months % | 18 to 24 months % | 24 to 36 months % | Over 36 months % |
|---|---|---|---|---|---|---|
| 2019-12-31 | 2325 | 40 | 15.7 | 12.4 | 14.4 | 17.5 |
| 2022-12-31 | 3910 | 29.9 | 13.2 | 12.5 | 9.4 | 35 |
| 2024-12-31 | 4283 | 20.6 | 9.2 | 8.7 | 16.8 | 44.7 |
| 2026-09-30 | 4625 | 20 | 8.6 | 7.2 | 11.6 | 52.6 |
A lender is not obliged to file a satisfaction when a loan is repaid, and many never do, so some of these charges belong to loans that have already gone. That inflates the level. It does not obviously explain the direction, since the filing habit has no known reason to have worsened since 2019 and the control series has not risen. We use the trend, not the level, as the evidence.
Recent loans are repaying faster, which is not what the aged book would suggest
24.9% of specialist development charges written in 2024 were filed as repaid within 18 months, up from 19.7% for the 2023 cohort and 16.1% for 2019.
Companies House charge register, Construction Capital analysis
View the chart data
| Cohort | Originated | Within 12m % | Within 18m % | Within 24m % |
|---|---|---|---|---|
| 2019 | 959 | 7.6 | 16.1 | 25.8 |
| 2020 | 584 | 9.9 | 20.5 | 31.2 |
| 2021 | 1265 | 10.9 | 19.8 | 33 |
| 2022 | 1230 | 12.7 | 22.4 | 34.7 |
| 2023 | 956 | 10.8 | 19.7 | 28.2 |
| 2024 | 938 | 13.9 | 24.9 | not yet observable |
The stock and the flow are telling different stories, and both are true. The stock measure asks how many old loans are still on the register. The flow measure asks how quickly each year's new loans get out. The 2023 cohort was the slow one: only 28.2% had repaid within 24 months, against 34.7% for 2022. The 2024 cohort has so far reversed that.
Our reading is that lenders tightened appraisal after 2022 and the loans they did write in 2024 were on shorter, more conservative programmes. We cannot prove that from the register, which shows when a charge was registered and released but nothing about the scheme. What the data does establish is that the difficulty sits with loans written earlier, not with the newest vintage.
A cohort can only be measured over a window it has fully lived through, so the 2024 cohort has no 24 month figure yet. Cohorts up to 2020 are reference only: the company register holds live companies only, so a scheme company that repaid and then dissolved has vanished from those years.
The specialist book is growing more slowly, but the wider development book is not
Over the four quarters to 2026 Q3, specialist lenders registered 962 charges and 700 were satisfied, a net addition of 262. The same measure was 626 in 2022.
Companies House charge register, Construction Capital analysis
View the chart data
| Quarter | New charges | Satisfied | Net | Provisional |
|---|---|---|---|---|
| 2021-Q1 | 319 | 135 | 184 | no |
| 2021-Q2 | 336 | 109 | 227 | no |
| 2021-Q3 | 321 | 111 | 210 | no |
| 2021-Q4 | 289 | 114 | 175 | no |
| 2022-Q1 | 288 | 159 | 129 | no |
| 2022-Q2 | 307 | 129 | 178 | no |
| 2022-Q3 | 356 | 142 | 214 | no |
| 2022-Q4 | 279 | 174 | 105 | no |
| 2023-Q1 | 237 | 163 | 74 | no |
| 2023-Q2 | 249 | 195 | 54 | no |
| 2023-Q3 | 213 | 197 | 16 | no |
| 2023-Q4 | 257 | 203 | 54 | no |
| 2024-Q1 | 176 | 166 | 10 | no |
| 2024-Q2 | 249 | 221 | 28 | no |
| 2024-Q3 | 237 | 179 | 58 | no |
| 2024-Q4 | 276 | 197 | 79 | no |
| 2025-Q1 | 176 | 181 | -5 | no |
| 2025-Q2 | 259 | 199 | 60 | no |
| 2025-Q3 | 210 | 185 | 25 | no |
| 2025-Q4 | 267 | 163 | 104 | no |
| 2026-Q1 | 242 | 206 | 36 | no |
| 2026-Q2 | 232 | 165 | 67 | no |
| 2026-Q3 | 221 | 166 | 55 | no |
A smaller net addition has two possible causes that mean opposite things: more repayments, or fewer new loans. For specialist lenders it is mainly the second. Compared with 2022, new charges are down by 268 a year and satisfactions up by 96.
That does not mean exit demand is about to ease, because specialist lenders are only part of development lending. Across the wider book, restricted to borrowers with a housebuilding code, new charges rose from 3,840 in 2022 to 4,413, and the net addition was 2,067, against 2,208. Lending appears to have moved across lender types rather than fallen away, and those loans will need exits of their own.
What this means if you are approaching practical completion
This is our opinion as finance arrangers, drawn from the evidence above. It is general commentary rather than advice on any scheme, and indicative structures vary by lender, asset and borrower.
- Start the exit conversation before the facility forces it. With selling times at 67 days to find a buyer before any legal work, a scheme with a few months left on its loan has little room. A development exit facility is easier to arrange with time in hand and sales evidence than with a default date approaching.
- Test the appraisal against today's selling speeds, not 2022's. Rebuild the cash flow with the current time to find a buyer, then add the legal process, and allow for buyers borrowing at higher mortgage rates. If the numbers need faster sales, consider sales period finance before you need it rather than after.
- Decide early whether a block sale is a real option. Company and investor buyers exist, but sales in this category show a much lower median than standard sales, so stress the debt against a lower price. If it does not repay, the plan needs another route, such as finance against unsold units.
- Treat a part-complete scheme as a different conversation. Where the build is unfinished, the funding gap and the exit need to be solved together. Part-complete development exit sets out what lenders generally want to see.
- Bring evidence, not hope. Lenders usually size an exit on what the units will achieve and how fast they will sell. Reservations, local comparables and a realistic sales plan tend to do more for your terms than any market average, including the ones in this report.
We would turn more cautious if the 2024 cohort's repayment rate falls back as it ages, if new development lending keeps outpacing exit-type lending into 2027, or if higher mortgage rates keep sales agreed below the previous year into next spring. We would turn less cautious if the count of loans past their typical term starts to fall.
How this report was built
Every figure is calculated from the sources below when the page is built. Nothing is estimated, and cuts below the minimum sample are shown as suppressed.
- Charges. Companies House mortgage and charge filings. A charge is security registered by a lender. It shows a loan was taken and when it was repaid, but not how much was lent, because filings have recorded no amounts since 2013. Registered to 30 October 2026, satisfactions filed to 2 November 2026.
- Lender populations. Specialist means charges held by a curated set of pure-play development lenders (10,127 charges). Development-like adds charges from development-active mixed lenders where the borrower carries a housebuilding code (SIC 41100 or 41202), the charge particulars use development terms, or a newly formed property company with no buy-to-let code is the borrower (66,728 charges in total). These mixed lenders also write buy-to-let books, so this wider measure is noisier. Location and exit-type figures use the strict version, with housebuilding codes only. Lenders are never named.
- Satisfactions. Filing is not compulsory, so every repaid figure is a floor and every aged figure is a ceiling. A quarter-end is provisional until 30 days after the data cut-off, which our tests between successive data vintages support: charge filings arrive within about 20 days, and figures at that age did not move materially when re-run three months later.
- Survivorship. The company register holds live companies only, so older cohorts are missing scheme companies since dissolved. Cohorts up to 2020 are reference only.
- Sold data. HM Land Registry Price Paid Data, England and Wales, detached, semi-detached, terraced and flats. New-build figures stop at 2025 Q3 because registrations lag completions by a year or more.
- Two new-build counts. National series in this report count every price paid category, giving 80,203 sales in the four quarters to 2025 Q3. Location reports and rankings count Category A only, giving 67,091 in a near-identical window. Most of the difference is Category B, which the buyer mix analysis measures directly.
- Regions. Charges are placed by property postcode where one can be identified: 43.6% of the specialist book and 65.3% of the development-like book. England and Wales only.
- Suppression. Shares and medians with fewer than 30 observations are withheld rather than estimated.
- Planning. Approved residential applications from council portals, deduplicated by authority and reference. Approvals lead completions by two to four years, and coverage differs by region, so totals are context and are not ranked.
- Selling time. Rightmove House Price Index, entered from published releases. It covers all listings, not new-build alone.
Contains HM Land Registry data © Crown copyright and database right 2026, licensed under the Open Government Licence v3.0. Companies House data is used under its public register terms. You are welcome to cite or chart these figures with a link to this page.
This report is general market commentary. It is not financial, legal, tax or investment advice, and it is not an offer of finance. DevExit is a finance arranger and introducer, not a lender.
Data vintage: Charges to 2026-10-30; satisfactions to 2026-11-02; Land Registry to 2026-09-30. Published 2026-11-03.