MHCLG’s building-control series estimated 35,910 new-build housing starts in England in the second quarter of 2026, 20% more than in the same quarter a year earlier. Estimated completions were 35,800, only 1% higher year on year and 3% lower than in the previous quarter. Meanwhile, the latest published annual measure of housing supply showed 208,600 net additional dwellings in 2024–25, 6% fewer than the year before and the third consecutive annual decline.
Those figures are not contradictory. They are measuring different things, at different stages, with different coverage and different publication timetables. A start records activity near the beginning of construction; a completion records a new-build dwelling reaching the end of it; net additions measure the wider change in England’s housing stock after new builds, conversions, changes of use, demolitions and other gains and losses have been taken into account.
The distinction becomes more important when the numbers are used to make bigger claims. The latest MHCLG indicators of new supply show starts rising strongly, but MHCLG also warns that recent starts have been affected by a sharp increase in units reported through the Building Safety Regulator. The annual net-additions series is more comprehensive, but it arrives much later. Interim EPC-based estimates are quicker, but less certain. Planning-permission figures sit further upstream again and are not a count of homes that will necessarily be built.
The useful question is therefore not simply whether “housebuilding is up” or “housing supply is down”. It is which measure is being quoted, which geography it covers, what period it refers to and how complete the underlying data are. This article deals with that measurement problem. The separate question of why Britain is not building enough homes belongs to the causes of non-delivery rather than to the statistics used to describe it.
Four measures, four different questions
| Measure | What it actually records | What it tells you | What it does not tell you |
|---|---|---|---|
| Planning permissions | Homes within residential schemes recorded as receiving planning permission; the commonly quoted England units series is currently compiled from Glenigan development data | The scale and direction of the potential future development pipeline | How many consented homes remain unbuilt, whether the schemes are viable, or when construction will begin |
| Housing starts | New-build dwellings recorded through building-control systems when foundation work begins | A timely indicator of construction activity and potential future completions | A complete census of every home being built, or a guarantee that every recorded start will complete |
| Completions | New-build dwellings recorded as ready for occupation or receiving a completion certificate | A timely indicator of new-build homes reaching the end of construction | Conversions, changes of use, demolitions or the full change in the housing stock |
| Net additional dwellings | New builds plus conversions, changes of use and other gains, less demolitions and other losses | England’s most comprehensive annual measure of how much the dwelling stock increased | A timely measure of current construction activity, or the source of the increase without looking at the components |
The definitions matter because even apparently simple words can hide methodological detail. ONS’s UK housebuilding methodology generally counts a dwelling as started when foundation work begins and as completed when it is ready for occupation or a completion certificate is issued. In a block of flats, however, all the dwellings in the block are counted as started when foundation work on that block begins. One large apartment scheme can therefore move a quarterly starts figure by hundreds of homes at once, even though those homes are nowhere near occupation.
Just as importantly, these four series should not be treated as a matched statistical funnel. A permission recorded this year is not necessarily the same dwelling counted as a start in the next period, and a start recorded in one quarter may complete months or years later. Net additions also include homes created by conversion or change of use that may never appear as conventional new-build starts, while demolitions reduce the total. Dividing permissions by starts, or starts by completions, and calling the result a “conversion rate” can therefore create a number that looks precise without measuring a coherent cohort.
Why net additions is the headline measure for England
For England, MHCLG describes annual net additional dwellings as its primary and most comprehensive measure of additional housing supply. The latest published annual figure is 208,600 net additions in 2024–25, 6% below the 221,410 recorded in 2023–24. The series had already edged down from 234,460 in 2021–22 to 234,290 in 2022–23 before falling more sharply in the following two years, so 2024–25 marked a third consecutive annual decline.
Net additions are broader than new-build completions because they measure what happened to the housing stock as a whole. In 2024–25, the total included 190,600 new-build homes, 17,710 gains from changes of use, 3,850 from conversions between houses and flats and 1,080 other gains, offset by 4,630 demolitions. A year with strong new-build completions can therefore produce a weaker net-additions figure if other losses rise, while conversions can add homes even when conventional new-build activity is subdued.
The series is also slower because it is assembled from more detailed sources. MHCLG’s technical notes show that 291 of England’s 296 local authorities supplied figures for 2024–25, with values imputed for the remaining five. The statistics are Accredited Official Statistics, but the latest year’s figure is still provisional and subject to scheduled revision; “more comprehensive” does not mean “incapable of changing”.
To bridge the gap between annual releases, MHCLG uses new-dwelling Energy Performance Certificate lodgements, adjusted for estimated demolitions, to produce more recent estimates of net additions. This produces a useful near-real-time guide, but the department is explicit that the estimate is less certain than the later annual series. The latest estimate puts 2025–26 at about 196,900 net additional dwellings. A useful reminder of the difference between the two measures came a year earlier: the subsequently published annual 2024–25 figure of 208,600 was 12,000 homes, or 6%, higher than the EPC-based estimate previously published for the same year.
The distinction is important whenever a live housing target tracker is quoted. The annual net-additions figure and an EPC-based estimate may both be described in shorthand as “homes delivered”, but they do not have the same evidential status.
What the current England data actually shows
The latest quarterly picture demonstrates how easily the measures can diverge. MHCLG estimated 35,910 starts in Q2 2026, up 6% on Q1 and 20% on Q2 2025, while 35,800 completions were 3% lower than in the previous quarter and only 1% higher year on year. Over the rolling year to June, starts were estimated at 136,330, 15% higher than the preceding 12 months, while completions were 143,770, unchanged year on year.
Even that apparent recovery in starts requires care. MHCLG says the higher starts recorded in Q4 2025, Q1 2026 and Q2 2026 are in part due to the Building Safety Regulator reporting far more units than before: 7,089 starts in Q4 2025, 6,276 in Q1 2026 and 5,758 in Q2, compared with a previous quarterly maximum of 1,206 in Q3 2025. The 20% year-on-year increase therefore cannot be read as a pure measure of underlying market acceleration. The contribution from BSR-reported starts has changed materially, and the headline series does not isolate how much of the movement reflects construction momentum and how much reflects the changed reporting contribution.
There is a broader coverage issue as well. MHCLG estimates that its building-control sourced data currently covers about 80% of housebuilding in England. The department’s technical notes for the Q2 release describe the series as a leading indicator of supply rather than the most comprehensive count. Recent history also contains a regulatory distortion: Q2 2023 saw an unusually high peak in starts, with developers thought to have brought some activity forward ahead of changes to building regulations. Comparisons using that peak need particular care.
Quarterly numbers can change after publication too. MHCLG leaves old releases as they appeared at the time, while its live tables incorporate later revisions when improved information arrives. Two articles written months apart can therefore quote different figures for the same quarter and both accurately reflect the official source available to them at the time.
The government’s 1.5 million-home ambition adds another layer. Because the full annual net-additions series currently reaches only to March 2025, progress since then is estimated using EPC data. MHCLG currently puts net additions from the start of Parliament on 9 July 2024 to 20 September 2026 at 437,900. BCIS’s 25 September analysis notes that this is about 29% of the 1.5 million target after roughly 44% of the five-year parliamentary period had elapsed.
Useful though that comparison is, it is not a live counter based wholly on completed annual returns. Part of the 437,900 is drawn from EPC-based estimates that can change when the later annual statistics replace them. The target tracker is therefore best read as the government’s current estimate of progress, not as a permanently settled count.
Why starts, completions and permissions can move in different directions
Starts and completions are related, but the time between them is not fixed. MHCLG says the journey from start to completion can take from a few months to several years, depending on the development. A single house, a phased estate and a high-rise apartment scheme do not move through construction on interchangeable timetables, and some recorded starts never complete at all.
A rise in starts is therefore a leading signal, not evidence that completed supply has already increased. Today’s completions reflect a mixture of starts made under earlier financing conditions, regulations, construction costs and market circumstances. The lag alone can make starts strengthen while completions weaken without any contradiction.
Planning permission sits further upstream again. The latest planning-applications statistics for England show about 212,000 homes granted permission in the year to June 2026, 12% fewer than a year earlier. Those unit figures come from Glenigan data and are provisional management information rather than Accredited Official Statistics; projects are revised and duplicates removed as far as possible as better information becomes available.
More importantly, 212,000 is not the stock of “homes with permission waiting to be built”. It is a flow of units receiving permission over a particular 12-month period. Individual schemes can be amended, superseded, re-permitted or never implemented, and the permissions series cannot be divided mechanically by starts or completions to infer how much of the pipeline has converted.
The reasons why a consent does or does not become a start are covered separately in PFP’s analysis of what actually stops new housing getting built. For this article, the narrower point is that permission, start and completion statistics describe different events and cannot substitute for one another.
England is not the UK — and the national series are not interchangeable
Housing coverage frequently shifts between “England”, “Britain” and “the UK” even though the underlying statistics do not share one perfectly uniform methodology. ONS compiles UK housebuilding data from the systems used in England, Scotland, Wales and Northern Ireland, allowing broad comparison while also documenting differences between them.
England’s net-additions series is an England measure, and the 1.5 million-home ambition discussed here is an England target. Neither should be presented as a UK total. ONS also notes differences in the recording of starts: in Northern Ireland the start date is based on the first building-control inspection, whereas the standard definition elsewhere is tied to foundation work.
Scotland publishes its own housebuilding statistics. The Scottish Government’s September 2026 release reported 17,057 new-build completions and 15,543 starts in the year to June 2026, with completions down 11% and starts down 2% year on year. Scotland also uses broader measures of housing supply within its own statistical system.
The lesson is methodological rather than competitive. An English net-additions figure, a Scottish new-housing-supply figure and a UK housebuilding starts figure can all be useful, but they should not be placed side by side as though they are three differently labelled versions of the same statistic.
Reading the next housing-supply headline
Housing statistics become much easier to interpret once the reader stops looking for one definitive live number. Planning permissions show potential future development. Starts offer a timely but incomplete signal of construction beginning. Completions show new-build homes reaching the end of construction. Net additions ask the broader annual question of how much England’s dwelling stock increased after gains and losses from several sources were counted.
The next step is to ask how the figure was produced. Is it a building-control estimate or the wider annual net-additions series? Is it an EPC-based interim estimate? Has the reporting coverage changed? Is the number still provisional? Has a later live table revised it? Does the geography say England or the UK? Is the comparison being distorted by an unusual regulatory deadline or by a large block of flats entering the starts count together?
A headline saying “starts are up 20%” can therefore coexist with another saying annual housing supply has fallen for a third year. Neither statement necessarily tells the whole story, but neither needs to be wrong. They may simply be describing different stages, different periods and different statistical systems.
For developers, investors, agents and anyone trying to judge the direction of housing delivery, the useful question is not merely whether the latest number went up or down. It is which number moved, what exactly it counts, how complete it is, what geography it covers and whether it can legitimately be compared with the number beside it. Once those questions are asked, much of the apparent contradiction disappears.
This article provides general information about how housing-supply statistics are measured and interpreted. It is not a forecast of future housing delivery. Housing data are subject to differing methodologies, coverage limitations and subsequent revisions, and provisional estimates may be replaced or revised by later official statistics.
