The Article 4 Wave: How England's Councils Are Closing the HMO Loophole

What is the Article 4 wave?

On Monday 17 August 2026, County Durham's countywide Article 4 direction came into force. One of England's largest local planning authorities, after a consultation that drew more than 1,400 responses running roughly 80% in support, removed permitted development rights for small HMO conversions across its entire area. From that day, converting a family home into a small house in multiple occupation anywhere in the county needs planning permission.

Durham is not an outlier. It is the largest entry so far in a wave that has been building for two years. The national planning.data.gov.uk dataset, downloaded 7 August 2026, holds 62 HMO-related Article 4 directions across 32 English councils, and no central register tracks their status, so PlanningLens built one. Every direction named in this briefing is verified against council primary sources: the notices, cabinet papers and confirmation decisions the councils themselves published. The full register, with confirmation status for each direction, is Table 4.

The timeline above is the shape of the story. A first generation of directions arrived quietly between 2011 and 2017: Bristol, Exeter, York, Barking & Dagenham, Plymouth, Enfield, Barnet, Havering, Stevenage. A second, smaller London cluster followed: Tower Hamlets in 2021, Brent in 2022, Hounslow in 2023. Then, from mid-2025, the wave: fifteen directions in force, pending or confirmed with future starts inside eighteen months, from Greater Manchester to Kent, with two more councils, Bury and St Helens, already in the pipeline behind them.

Why are councils moving now?

The wave has three distinct shapes, and every council named below is verified to its own published record.

Greater Manchester is closing, borough by borough

Trafford went first: an immediate, borough-wide direction in December 2017. Salford covered its inner areas in 2018 and later expanded. Manchester restricted its inner neighbourhoods and expanded the area to prevent spread. Then the pace changed: Bolton (immediate, June 2025), Wigan (in force 31 August 2025), Tameside (immediate, October 2025), Oldham (January 2026), and Bury, whose immediate direction was made on 16 July 2026 with its confirmation consultation closing 27 August. Eight of Greater Manchester's ten boroughs now have HMO Article 4 coverage. By Bury's own January 2026 committee survey, Rochdale and Stockport are the holdouts.

The displacement arms race, in the councils' own words

Why do partial directions become borough-wide ones? Bolton's cabinet report states the mechanism plainly: evidence from Blackburn with Darwen and Salford showed that small-area directions displace HMO growth to the streets outside them, and both councils subsequently expanded their coverage. Partial directions leak; councils go borough-wide; growth is pushed into neighbouring boroughs; the neighbours act. The Greater Manchester sequence is this mechanism executing in real time.

It is now attested three times over, in three regions, in councils' own papers. Merton's consultation feedback records pressure displaced into its uncontrolled western wards after its 2022 seven-ward direction. And Hyndburn, whose nine-ward direction came into force on 15 March 2026, launched an immediate borough-wide extension the same month under urgency procedures, explicitly to beat a run of conversions into the uncovered wards. Hyndburn's original direction was covered by LandlordZone; the extension is the sequel. North West, London, Lancashire: the same displacement logic, written down by the councils experiencing it.

Not every part of the wave is an arms race. Central Lancashire (Preston, Chorley and South Ribble) built a joint Article 4 evidence base under their shared local plan in February 2025, and all three now have directions. Where Greater Manchester is competitive displacement, Lancashire is coordination. Two different engines, one wave.

The wave is accepting compensation risk

An Article 4 direction made with less than twelve months' notice exposes a council to compensation claims from landowners with stranded plans. The wave's most striking feature is how many councils are accepting that risk to close the loophole immediately: Bolton (June 2025), Chorley (September 2025), Rossendale (September 2025), Tameside (October 2025), Dartford's north-of-A2 area (December 2025), Hillingdon (December 2025), South Ribble (March 2026), Merton's thirteen wards (March 2026, confirmation pending), Hyndburn's extension (March 2026, confirmation window open) and Bury (July 2026) were all immediate. Trade press called Bolton's no-notice direction a wake-up call, with conversions worth £200,000–250,000 stranded mid-build.

The counterpoint deserves equal weight: Merton reports zero compensation claims within twelve months of its 2022 seven-ward immediate direction. The risk is real in law and sometimes prices at zero in practice.

And there is a quieter fourth lane: councils taking the non-immediate route and surviving consultation cleanly, with dated future starts: Wigan (31 August 2025), North Lincolnshire's Scunthorpe wards (26 November 2026, on roughly 90% consultation support), Warrington (borough-wide, by end September 2026, per the council), and St Helens, whose Reform-led council has, per reporting in LandlordZone, voted to bring in a borough-wide non-immediate direction. Voted, not yet made; no notice yet exists.

The wave's outer edge

Two councils mark how far the logic now reaches. North Warwickshire's borough-wide direction, in force 2 March 2026, is the pre-emptive case: the borough has 0.44 licensed HMOs per 1,000 dwellings, the lowest rate in Warwickshire and around 0.1% of its housing stock, and acted anyway. LandlordZone covered its consultation in November 2025; the direction is now the epilogue. The wave includes councils moving before any concentration exists to prevent. And at the other end of the timeline sits Milton Keynes, an early adopter with two directions from 2010–11, and the council that took the government to judicial review over its HMO planning rules. The instrument has been contested from the start; what is new is how many councils now want it.

What does the new NPPF mean for the wave?

On Monday 17 August, hours after County Durham's countywide direction came into force, the government published the first comprehensive rewrite of the National Planning Policy Framework since 2012. For the councils behind the Article 4 wave, it delivers a validation and a problem in the same paragraph.

The validation: for the first time, national policy names the issue this wave exists to address. Policy DM10 states that Article 4 directions should be limited to situations where they are necessary to protect local amenity or wellbeing, and its examples now include preventing an over-concentration of uses affecting quality of life "such as small Houses in Multiple Occupation". The rationale that 32 councils have written into cabinet papers since 2024 is now national policy's own example.

The problem: the same policy requires every direction to be based on robust evidence and to apply to "the smallest area required to mitigate the evidenced harm". That test points squarely against the form the wave has actually taken. Of the twelve directions confirmed since January 2025 whose coverage PlanningLens has verified, eight are borough-wide or countywide, and the direction of travel is one-way: Hillingdon and Spelthorne both superseded earlier ward-level directions with borough-wide ones, and Hyndburn's borough-wide extension is in its confirmation window now.

And the councils' own evidence explains why they went big anyway. Bolton's cabinet papers cite the experience of Blackburn with Darwen and Salford: when a direction covers a small area, HMO growth is displaced to the streets just outside it, so partial directions get expanded, growth gets pushed into neighbouring boroughs, and neighbours act. The displacement evidence and the smallest-area test pull in opposite directions, and every future direction will have to argue its extent in the gap between them.

Two boundaries on what this means, stated plainly. Existing confirmed directions are not retrospectively tested against the new framework: Durham's direction was made and confirmed under the old one, and came into force hours before the text that would have interrogated its extent. And the tension is prospective: it constrains the wave's next members, not its current ones. Whether the smallest-area test slows the march toward countywide coverage, or simply lengthens the evidence documents that accompany it, is the question the next twelve months of directions will answer.

Does Article 4 change decisions?

An Article 4 direction changes what needs permission. Whether it changes what happens to the applications that follow is an empirical question, and the honest answer from the decision record is: sometimes, partially, and less mechanically than either side of the argument assumes.

Before and after, in three boroughs

For the three London boroughs whose directions took force recently enough to compare like periods (three years either side, direction year and partial 2026 excluded), the record shows:

Windows: before = 3 full years preceding the direction year; after = up to 3 full years following, direction year and 2026 excluded. Hounslow's after-window is two years (2024–25). Source: a4_before_after_cases_fresh.csv, recompute of 17 Aug 2026.

Three boroughs, three different answers. Tower Hamlets saw volume rise by half and refusals climb six points. Brent, the largest base in the comparison, is a null result: volume flat, refusal rate marginally down. Hounslow's refusal rate jumped ten points on a short after-window. The Brent null is reported with the same prominence as the Hounslow rise, because a tool that produces three different outcomes in three neighbouring boroughs is not a switch. It is an instrument whose effect depends on what each council does with it.

The long-established cohort

For councils whose directions have been in force for eight years or more, the question shifts: what does the steady state look like? The answer depends less on having a direction than on how much of the council it covers.

All dates and coverage verified to council sources. Basingstoke & Deane excluded: coverage unverified on the platform record and a five-application base. Small bases (Stevenage, 9 decided) are shown, not hidden. Hillingdon's borough-wide direction took force 11 December 2025 (confirmed under seal 21 May 2026); its 2024–25 rate is therefore a rate under two-ward coverage. Source: a4_long_cohort_fresh.csv, recompute of 17 Aug 2026.

The Enfield correction

An earlier PlanningLens analysis flagged Enfield's 657% growth in decided HMO applications, alongside a softening refusal rate, as a possible "Article 4 signature", offered explicitly as a hypothesis. We then checked it against the direction record. Enfield's direction is borough-wide and has been in force since 22 October 2013: it predates the 2021–25 surge by eight years and cannot explain it. The hypothesis is rejected. The surge remains unexplained; licensing changes and enforcement activity are candidate explanations, and they are named here as candidates only. We flagged it, we tested it, and we are reporting the negative, for the same reason the Brent null gets its own paragraph above.

What happens when you're refused?

An Article 4 direction determines whether you must apply. The refusal statistics above determine your odds. The appeals data determines what refusal costs. In the highest-refusing boroughs, the answer is that the second roll of the dice is loaded too.

Nationally, appealing a change-of-use refusal adds a median 238 days to the process (interquartile range 168 to 362 days, across 11,053 decided change-of-use appeals to December 2025). The wait peaked in 2023 at a median 285 days and fell to 188 by 2025. Across all decided change-of-use appeals, 31.2% succeed: 2,373 allowed against 5,242 dismissed. Roughly a national one-in-three.

Now put that baseline against the boroughs that refuse the most. Where PlanningLens's appeal-matching gate passes for a council, HMO-specific figures are given; everywhere else the figures are all-change-of-use casework and labelled as such.

¹ All refusal rates are 2024–25 decided HMO applications. Where the 17 August recompute covers a council (Barking & Dagenham, Havering, Barnet), the recomputed rate is used; marked rows carry the tranche-1 join rate. Appeal success = allowed / (allowed + dismissed); enforcement appeals excluded; councils failing the reference-matching gate are excluded from HMO-specific figures rather than estimated. National baseline (31.2%) is all-change-of-use casework and is compared like-for-like in the text. Sources: PINS appeals release July 2026; q5_hmo_outcomes.csv, q5_hmo_durations.csv.

Barking & Dagenham is the near-total shutdown: 94% of HMO applications refused, and of the fifteen HMO refusals taken to appeal and decided, one succeeded. One in fifteen, against a national change-of-use one in three. On the like-for-like basis its all-change-of-use appeal record reads 7.1% (3 of 42) against the 31.2% national baseline, 4.4 times worse, so the HMO-specific figure is not an artefact of narrowing the sample.

Barnet is the finding this data surfaces for the first time. Its headline refusal rate, 61.9%, is mid-table for London. But its HMO appeal record is indistinguishable from Barking & Dagenham's: three allowed of forty-six decided, 6.5%, on the largest HMO appeal base in the matched set. An applicant refused in Barnet faces better-than-even odds of refusal and then a one-in-fifteen appeal: a double bind that the refusal rate alone entirely conceals.

One honesty note on Havering, because the dramatic version of its story does not survive the better data. On all-change-of-use casework, Havering's appeal median is 464.5 days, fifth-slowest in England, and the origin of the line "refused, then a fifteen-month wait". Its HMO-specific appeals, however, resolve in a median 170 days (n=17). The double bind in Havering is the 93.2% refusal rate paired with a one-in-eight appeal, not the wait. The fifteen-month figure belongs to its wider change-of-use caseload, and that is where this briefing leaves it.

The delay extremes run elsewhere: among matched councils, Ealing's change-of-use appeals run a median 475.5 days and Brent's 429; the borough where Article 4 produced a null effect on refusals is still the second-slowest place in the set to appeal one.

Who decides, and does party control matter?

The wave invites a partisan reading. The decision data declines to supply one.

Labour-run councils occupy both extremes of the refusal table: Barking & Dagenham (94.0%) was won 51-0 by Labour in 2022, and Plymouth (3.0%) and Mansfield (every decided application approved) were Labour-controlled through the window too. No simple partisan story survives that spread. It rhymes with the finding that runs through this whole briefing, that Article 4 is a tool whose outcome depends on deployment, not a party platform.

On Reform UK, the record permits exactly one statement: there is no Reform-attributable decision data in the window. Reform's only council among those examined here is Havering, gained in May 2026, after the 2024–25 window closed. Every decision in these tables predates Reform control anywhere. What a Reform administration does with a 93%-refusing borough, and what St Helens's Reform-led council does with the borough-wide direction it has voted for, is a 2027 question, and PlanningLens will run this analysis again when there is data to run it on.

Two cautions frame anything stronger. First, the administrations that produced the 2024–25 numbers have half-dissolved: May 2026 turned Havering to Reform, Enfield and Exeter to no overall control, cut Hounslow's majority to one and left Barnet on a knife-edge. The window is one electoral cycle, and it is treated as one. Second, the mechanism is weaker than it looks: most HMO decisions are officer-delegated. Elected members act mainly through policy (Article 4 directions and concentration thresholds), not case-by-case. Control-to-refusal causality is weak by construction, and what this section reports is descriptive association plus the policy channel, which is exactly the channel the wave itself runs through.

Which directions are actually in force? The confirmation-status table

An immediate Article 4 direction is only in force today if it was confirmed inside its statutory window, and no central register records whether that happened. Platform date fields are unreliable in four documented ways (see methodology). So PlanningLens verified each direction below against council primary sources: the making notice, the confirmation decision, the cabinet paper. This table is, to our knowledge, the only published confirmation-status register for the HMO Article 4 wave.

² Chorley: month precision only; the council's 23 September resolution is a decision date, not a making date, and PlanningLens does not fake day precision it does not have. Durham's made and confirmed dates are deliberately blank: only the in-force date is verified. Dartford is two directions with distinct routes and start dates; showing it as one row would be wrong. Established-era rows carry in-force dates only; blank made/confirmed cells mean the date is not council-source verified, not that it does not exist. Future-start rows are confirmed directions with dated future commencements. The register covers directions taken through full status verification; further verified directions named in the text (the Greater Manchester early movers, North Warwickshire, Middlesbrough and others) join the register as their status checks complete. Verified against council primary sources throughout; register maintained by PlanningLens.

How this was put together, and what it cannot tell you

1. Direction dataset. planning.data.gov.uk article-4-direction dataset, downloaded 7 August 2026: 3,234 directions nationally, filtered to 62 HMO-related directions (57 dated) across 32 councils. Every council named in this briefing is additionally verified against council primary sources.
2. Decision data. PlanningLens planning-decision records, HMO applications identified from application text with false-positive filtering, on the same basis as Briefing #02.
3. Coverage is incomplete. The PlanningLens estate does not cover every English council, so no "Article 4 versus no Article 4" national comparison is made anywhere in this briefing.
4. Annual granularity. Before/after windows are built from full calendar years; sub-annual effects are invisible to this design.
5. The selection-effect confound. A direction changes which applications exist, not just how they are decided: Middlesbrough registered 16 lawful-development-certificate applications in four days around its February 2025 deadline, three on one street. Before/after refusal rates compare different populations of applications, and no causal claim in this briefing ignores that.
6. Coverage extent is a variable, not a footnote. Demonstrated six ways in the data: Hillingdon (2 wards, then borough-wide), Bristol (staged), Plymouth (city centre), Exeter (university wards), Fenland (Wisbech only), Middlesbrough (town-centre carve-out). Every table in this briefing carries a coverage column for this reason.
7. Platform dates are not print-grade. Four error types documented in the national dataset: wrong made-dates, January-1st placeholder artefacts, confirmation dates recorded as making dates, and cabinet-approval dates recorded as making dates. Every date printed here is council-source verified; where only month precision is defensible, only month precision is printed.
8. Confirmation status gates everything. An immediate direction lapses if not confirmed inside its statutory window. Table 4 is the register; pending confirmations (Merton, due 16 September 2026; Bury, window to January 2027; Hyndburn's extension) are labelled, not assumed.
9. Appeals. Planning Inspectorate casework to December 2025, change-of-use category, decided appeals; success = allowed/(allowed+dismissed), enforcement excluded. HMO-specific figures are given only where PlanningLens's reference-matching gate passes for a council; failing councils are excluded, not estimated. Bases under 10 are suppressed.
10. Political control. Web-verified per council for both the 2024–25 window and August 2026. Most HMO decisions are officer-delegated; members act mainly through policy, so party-control findings are reported as descriptive association only.