The HMO Squeeze

The squeeze

England's planning system is deciding nearly twice as many HMO applications as it was four years ago, and refusing a sharply rising share of them.

Across a stable panel of 144 English councils, decided applications for houses in multiple occupation rose from 1,848 in 2021 to 3,454 in 2025: an increase of 87%.¹ Refusals more than doubled over the same period, from 590 to 1,203. The approval rate has slipped from 68.1% in 2021 to 65.2% in 2025, with partial 2026 data running lower still at 63.2%.

The tightening is concentrated where it is most consequential: applications to create an HMO. Among these, the share refused rose from 32.3% in 2021 to 35.5% in 2025.

This is not an artefact of growing coverage. Across all 219 English councils in the PlanningLens estate, decided HMO applications rose 75% between 2020 (2,192) and 2025 (3,830). The stable panel, whose membership does not change, shows steeper growth than the expanding all-councils base. The surge is in the system, not in the sample.

The squeeze arrives at a moment when national policy depends on this exact asset class. The asylum accommodation contracts let by the Home Office were designed around dispersal accommodation, defined as longer-term self-catered housing such as flats and HMOs (Home Affairs Committee, The Home Office's management of asylum accommodation, HC 580, 2025). That accommodation is procured by three private suppliers, Clearsprings, Mears and Serco, under seven regional contracts (National Audit Office, The Home Office's asylum accommodation contracts, May 2025). And since 2022, every local authority area in England has been a dispersal area, with allocations proportionate to population (House of Commons Library, SN01909).

At the same time, the planning restriction on HMOs is spreading. No central register of Article 4 directions exists, but the National Residential Landlords Association estimates that 75 to 80 English local authorities now have directions removing permitted development rights for small HMOs, requiring planning permission where none would otherwise be needed.² Birmingham's city-wide direction, in force since 8 June 2020, is the most prominent documented case.

Demand for the asset class is national policy; supply of permissions is contracting at the local level.

Key findings

• Decided HMO applications rose 87% between 2021 and 2025 across a stable panel of 144 English councils (1,848 to 3,454), and refusals more than doubled, from 590 to 1,203.
• The approval rate held near 68% from 2021 to 2023, then fell three points across 2024 and 2025, with partial 2026 data running lower again at 63.2%. The squeeze is recent, and it is steepening.
• Refusal rates among citable councils span 4.8% (Lincoln) to 80.7% (Havering), and the five highest refusal rates among threshold-clearing councils all belong to London boroughs.
• Asylum dispersal and HMO activity concentrate in the same councils, and neither predicts the other. The link everyone assumes does not survive scrutiny, and this report says so.

What the planning record shows

The figures above come from a fixed panel of 144 English councils, each of which decided at least 50 HMO applications in every year from 2021 to 2025. The threshold is a stability criterion, not a selection: it measures the trend across councils with continuous, comparable activity, so no part of the rise can be produced by councils entering or leaving the data.

The 2020 row is shown for completeness; decisions that year were taken under pandemic disruption, and 2021 is the comparison base throughout this report.

The shape of the trend matters as much as its size. Volume growth is concentrated in the most recent years: decided applications were broadly flat to 2022, then rose in each of the next three years, adding nearly 1,500 decided applications between 2022 and 2025. The approval rate tells the same story from the other side: it held near 68% in every year from 2021 to 2023, then fell three points across 2024 and 2025, with partial 2026 lower again. The three-point fall translates into counts: had the 3,454 applications decided in 2025 been approved at the 2021 rate of 68.1%, roughly 100 more permissions would have been granted in 2025 alone. That is a backward-looking comparison, not a forecast. The squeeze is not a long, slow drift. It is recent, and it is steepening.

Within the overall caseload, the direction of the application matters. Applications to create an HMO, the conversion of existing housing into shared occupation, are refused at a higher and rising rate: 32.3% of decided to-HMO applications in 2021, 35.5% in 2025. This statistic uses the v2.1 direction classifier; the stricter v2 classifier, run as a cross-check, shows the same rise (31.4% to 35.7%). The methodology box sets out the difference.

One known gap understates the surge rather than inflating it. Liverpool (2,293 decided HMO applications) and Birmingham (993) currently hold decision records without usable dates, which makes them invisible to the trend figures. When those records are dated, the national rise can only grow. The figures published here are therefore conservative.

Where the squeeze bites

The national figures hide the mechanism. HMO activity is intensely localised, and so is the tightening: the same national squeeze produces opposite responses in different council chambers.

The league cut behind this section covers 178 English councils and 14,382 decided HMO applications across 2021 to 2025, on a council-year basis. Two disciplines govern what is cited from it. The table below shows the top 20 councils by decided volume, every one of which decided at least 236 applications over the period. Refusal rates for councils outside the table are cited only where the council decided at least 100 applications; 40 councils clear that bar.

The span is the story. Among the 40 councils clearing the citation threshold, refusal rates run from 4.8% in Lincoln (395 decided) to 80.7% in Havering (135 decided). Within the top-20 table itself the spread is nearly as wide: Lincoln again at the bottom, Waltham Forest at the top at 75.5%. These are councils processing the same kind of application under the same national framework, in the same five years, reaching opposite defaults.

The geography of the high end is not random. The volume leaders are not the refusers: Leeds, the busiest council in the table (904 decided), refused 21.6%, and Portsmouth (771 decided) refused 13.6%. The refusing end belongs to London. Five boroughs in the table refused more than half of everything they decided: Waltham Forest (75.5%), Croydon (65.7%), Brent (61.8%), Barnet (57.0%) and Haringey (51.7%). The five highest refusal rates among all threshold-clearing councils are likewise all London boroughs, a composition pattern that section 4 returns to.

One caveat governs how any map of this data should be read. Where no Article 4 direction is in force, converting a family home to a small HMO is permitted development and never enters the planning record. Low application counts in non-Article-4 councils therefore partly reflect the absence of a requirement to apply, not the absence of HMO activity. No central register of directions exists; the National Residential Landlords Association estimates 75 to 80 English local authorities have them, and Birmingham's city-wide direction, in force since June 2020, is the most prominent documented case.

The link everyone assumes

The most charged subject adjacent to this data is asylum dispersal. The position of this report is simple: we joined the data, and we report exactly how far the link goes and where it stops.

We joined twelve years of Home Office dispersal data to our planning records and tested the link everyone assumes. HMO activity and asylum dispersal concentrate in the same councils, and that is where the connection ends: changes in one do not predict changes in the other, in either direction, and high hotel populations do not predict refusal rates outside London.

What the data supports is co-location. Councils with more dispersal accommodation see more HMO applications: across the joined panel, the correlation between levels is 0.43, and between relative growth rates 0.39. Part of the levels association is simply council size. Normalised per 10,000 residents, the co-location weakens to 0.27 but remains positive. Both numbers are reported because the weaker one is the fairer one.

What the data does not support is sequence. Dispersal growth does not precede HMO application growth (correlation −0.09), and HMO application growth does not precede dispersal growth (−0.11). Neither moves first.

One reading is consistent with this pattern, and it is offered as interpretation, labelled as such: shared geography. The same cheap, convertible housing stock attracts both HMO landlords and dispersal procurement. The data shows them in the same places. It does not show one causing the other.

We also tested the claim a reader is most likely to bring to this report: that the places absorbing the most people refuse the most HMOs. It does not survive scrutiny. Pooled across England, councils in the highest dispersal tercile refuse 41% of HMO applications, but that figure is London composition, not a dispersal effect. Outside London the gradient vanishes: high-dispersal councils refuse 24.0%, councils with no dispersal accommodation 21.3%, and the middle tercile is lowest of all at 15.4%. Normalising by population does not rescue the claim; terciled by hotel residents per 10,000, the ex-London pattern is still flat (23.3% against 21.3%, middle tercile lowest at 18.3%). The counterexamples are not hard to find either: Rugby hosted 391 hotel residents while deciding 27 HMO applications in the window and refusing none of them. We are not publishing the claim, and we are saying so.