The Permissive Ten

The companion paper established that London runs more than thirty separate planning systems for small sites, and that where a borough lands on the outcome differential tracks the price of its housing. This follow-up turns that instrument on the ten boroughs that approve most generously relative to what their caseload should yield, and asks the question the price gradient invites. When a borough says yes more often than expected, is it because its plan says yes, because it has no choice, because of who runs it, or simply because of who is asking?

I will argue that ten superficially identical positive numbers are produced by at least four different mechanisms, and that only one of them is something another borough can decide to adopt.

I Why these ten: the boroughs that say yes

The differential is the borough's observed approval rate minus the rate London-average decision behaviour would have produced on that borough's own caseload. A positive number means more approvals than the mix predicts. It is not a measure of intent, and it does not control for unobserved application quality. The ten boroughs below are the ten largest positive residuals in the 4 June 2026 build. They span from Camden to Richmond on outcome, and from Bexley at £420,000 to Kensington & Chelsea at £1.1 million on price.

Bexley and Kensington & Chelsea sit within four points of each other on the residual and more than two-and-a-half times apart on median price. A single mechanism that ran through house prices would not place them together. The rest of this note separates the routes, because the routes are what an authority trying to learn from these boroughs actually needs, and three of the four cannot be adopted.

II What the number hides: permissiveness has a shape too

Disaggregating the borough residual to site type, with cells below thirty decisions suppressed, is the cut that made restriction legible at the other end of the table. Applied here it returns a different pattern. The permissive residuals are, with one exception, lifted across the types that have volume. Generosity in this group is a general posture more often than a typology-specific rule.

The exception is the cell that shows what a plan looks like when it is doing the work. Camden's conversions sit at +26 (n=89) and its demolish-and-rebuild at −7 (n=79). That split is in the adopted text: housing policies that support the subdivision of large houses into self-contained flats, and a climate-mitigation policy that subjects substantial demolition to a retention test. Where the plan is precise, the type cut is precise. The other nine rows are close to flat. Their plans are mostly not what is producing the residual.

III The plans mostly do not say yes

Read the adopted plans and the obvious model, that a positive residual flows from a permissive plan, survives in one of the ten. Several of the most generous boroughs carry policies that, read on their own, ought to suppress small-site approvals rather than raise them.

Kensington & Chelsea's housing policies include a resistance to amalgamation, a defence against the merging of flats into single large houses, and a design and heritage regime that gates almost every site. Westminster's City Plan encourages small-site delivery in principle, then caps new homes at 200 square metres, requires 20 per cent affordable floorspace on even sub-ten-unit schemes, and protects existing units against loss. Hammersmith & Fulham runs a standard, London Plan-conformant supply policy with heritage protection layered on top. None of the three reads as a permissive small-sites plan. All three sit well above expectation.

Southwark and Haringey make the same point from the other direction. Southwark Plan Policy P3 refuses the sub-division of a single dwelling of 130 square metres or less original net internal floorspace. Haringey carries a presumption against the loss of garden land and a family-housing protection zone. Both restrain small sites in the adopted text. Both still approve above expectation. Southwark's London Plan small-sites target is among the largest in the city, at 601 homes a year. A heavy small-sites obligation, and an administration prepared to grant and defend the borderline scheme, can outweigh a written brake. That is a structural reading, not a textual one.

Across 31 boroughs with a comparable Land Registry median, the residual and price move together (Pearson r ≈ +0.49 on price, +0.53 on log price). The slope is real and moderate. It is not a law. The boroughs that sit furthest above the line, Bexley, Sutton and Ealing, are the ones whose yes does not run through the price of their housing.

IV Four ways to say yes

Reading the ten plans against the ten residuals resolves the single positive number into four routes. The table assigns each borough to the mechanism that best accounts for its position. Where two operate, the table names the dominant one.

Permissive by plan. Only Bexley arrives at a positive residual by the route the ranking invites. Local Plan Policy DP2, adopted 26 April 2023, states that proposals for new dwellings on small brownfield sites, including infill and backland, will generally be encouraged within the borough's Sustainable Development Locations, and permitted outside those locations subject to the spatial strategy. Windfall is built into the housing trajectory. It approves because it has decided to, and written the decision down where an applicant, an officer and an Inspector can all read it. Camden belongs here with a qualification: its plan plainly drives the type split, so it is a borough with a written posture rather than a generally permissive one.

Permissive by structure. Ealing, Southwark, Haringey and Sutton are generous because of arithmetic and administration rather than encouragement. Each carries one of London's heavier small-sites obligations (Southwark among the top three boroughs by target, Sutton reliant on small sites for 64 per cent of its housing supply) and each runs a delivery-minded administration inclined to grant and to defend. Two of them, Southwark and Haringey, do this in spite of adopted policies that restrain subdivision and backland, which is the clearest possible demonstration that their generosity is a posture rather than a policy. Sutton's case is the structural one in its purest form: an ageing 2018 plan that loses weight against the small-sites presumption with every passing year, in a borough that cannot meet its numbers any other way. It says yes because it must.

Permissive by wealth. Richmond, Kensington & Chelsea, Hammersmith & Fulham and Westminster are the boroughs whose plans least explain their residuals. Once the plan has been read, what remains is the application. The next section takes that mechanism up, because it is the one that matters most for reading the table and travels least as a model.

V The wealth mechanism, and why it cannot be copied

The differential already controls for the observable mix: site type, transport accessibility, conservation coverage, unit count. It cannot control for the quality of the application itself, because the dataset does not record the architect, the planning consultant, the heritage statement, the daylight model, or the capacity to absorb a section 106 contribution on a small scheme. In the dearest boroughs the same site arrives in front of an officer as a better-prepared proposal, because the residual on the land can pay for that preparation. A plan that gates on design quality is not a permissive plan. It behaves like one when the applications that arrive can clear the gate.

That is why the four wealthy boroughs over-approve under plans that read as neutral or restraining. Kensington & Chelsea's amalgamation control and Westminster's floorspace cap and affordable-housing charge are real. They bite hardest on the under-resourced application, which is the application that does not arrive. The schemes that do arrive are assembled to pass, and they pass. The residual records the result as generosity. It is more accurately a borough whose applicants can meet a demanding plan on its own terms.

The consequence for any borough trying to learn from these examples is decisive: the wealth mechanism cannot be adopted. A council cannot vote to receive better-prepared applications. It can only change its plan, change its posture, or wait for its plan to age, which is to say it can choose Bexley's route, or Sutton's, or Ealing's, but not Kensington & Chelsea's. The most generous boroughs in London are, for the most part, the least useful as models, because the thing producing their generosity is not in their gift. The useful models are the cheap ones.

VI The appeal record at the permissive end

A refusal in this system is close to terminal. About six in a hundred are appealed. Of those that are, the Inspectorate upholds the borough more often than not. The overturn rate is not flat. In the matched small-site appeal cohort the hardest boroughs are overturned at about 16 per cent of decided appeals and the most permissive at about 24 per cent, with the association running in the direction that first looks backwards (r = +0.40 between the difficulty gap and the overturn rate; London overturn 21.0 per cent, 458 decided appeals). Richmond is overturned at 36.7 per cent of its decided cases. That is not evidence that permissive boroughs decide badly. A borough that approves most of what it receives refuses only the schemes it is least sure of, and those are the schemes an Inspector is most likely to allow.

Two consequences follow. The permissive residuals are granted at first instance, by the borough, not imposed on review. And the appeal rate is a poor measure of stance: a high overturn rate at this end of the table means the borough rarely refuses, not that it refuses badly. The appeal corpus describes the texture of the decisions that survive. Only the upstream plan establishes the reason behind them.

VII The spine: Richmond versus Bexley

Richmond and Bexley sit close together at the head of the table, +21.8 and +18.9, and a quarter of a million pounds apart on median price. Bexley says yes in writing, through Policy DP2, at a London-cheap price, on a caseload that is lifted across conversions (n=151) and demolish-and-rebuild (n=86). Richmond's 2025 Local Plan adopts the small-sites duty, then gates it on character and garden protection, in a borough wealthy enough that its applicants clear those gates as a matter of course. Take the wealth out of Richmond and the character gate would start refusing schemes. Take the wealth out of Bexley and the plan would still say yes, because the plan is what says it.

For a hard, cheap borough asking what a positive residual would cost, Havering sits at almost the same price as Bexley and forty points below it on the differential. The instructive neighbour is Bexley, not Richmond. Richmond's example asks a borough to become wealthy. Bexley's asks it to write an encouraging small-sites policy, locate it, and mean it.

VIII Which yeses are durable

The four routes also differ in how long they last. Bexley's residual is the most durable because it is the most deliberate: it lives in an adopted plan and will outlast any given administration or applicant. The structural residuals are contingent. Ealing, Southwark, Haringey and Sutton are generous while their targets are high, their plans are old or silent, and their administrations are willing. Change any of the three and the posture can reverse. Southwark and Haringey both changed political control in May 2026, after this window. A residual that rests on disposition is the kind a change of disposition can withdraw. The wealth residuals are durable and inimitable: they will persist as long as the applicants are rich, which is indefinitely, and they offer nothing a poorer borough can use.

Read for lessons rather than for ranking, the table inverts. The cheapest borough of the ten is the only one whose generosity another borough could choose to reproduce. The famous and expensive ones are the least useful, because their advantage is the one thing a plan cannot legislate.

IX Conclusion

Ten identical-looking positive numbers are produced by four different mechanisms: a written plan, a structural obligation, an administrative posture, and the wealth of the people applying. Only the first is available to a borough on request. The differential found the generosity; the type cut showed that it is broad rather than written; the plan audit confirmed that, for most of these boroughs, it is not in the plan at all. The boroughs that say yes are not, in the main, the boroughs with permissive plans. They are the boroughs that can afford to behave as though they had one, and the single exception, which actually has one, is the cheapest of them all, and the only one worth copying.

Sources & confidence

Outcome differentials and type-disaggregated cells: London small-sites planning dataset (LSSPD), build 4 June 2026 (borough_differential.csv, type_differential_cells.csv). Prices: borough median, Land Registry (national_ps2_wealth_gradient.csv); Pearson r ≈ +0.49 on price and +0.53 on log price across the 31 boroughs with comparable medians. Appeal overturn: LSSPD matched-appeal cohort, 458 decided small-site appeals; London overturn 21.0%, permissive boroughs about 24%, r ≈ +0.40 with the difficulty gap. Plan audit: Bexley Local Plan (adopted 26 April 2023, Policy DP2, including the main-modification wording that locates the encouragement in Sustainable Development Locations); Richmond Local Plan (adopted 7 October 2025, Policy 16); Kensington & Chelsea Local Plan (adopted 24 July 2024, Policies HO1/HO2); Westminster City Plan 2019–2040 (partial review adopted January 2026, Policies 12/13); Hammersmith & Fulham Local Plan (adopted 28 February 2018); Camden Local Plan (adopted 3 July 2017, Policies H3/H7/CC1); Ealing development-plan suite (2012–2013, new plan at examination); Southwark Plan 2022 (adopted 23 February 2022, Policy P3); Haringey DPDs (adopted 24 July 2017); Sutton Local Plan 2018. Small-sites share of target: hdt_small_sites.csv (Sutton 64%).

What this can and cannot tell you. The descriptive claims are firm: these ten boroughs approve above their caseload-expected rate, their generosity is broad across site types with volume, and their adopted plans range from actively encouraging (Bexley, inside its Sustainable Development Locations) to actively restraining (Kensington & Chelsea, Westminster, Southwark P3). The mechanism attribution is an interpretation, not a measurement. The "wealth" reading is an inference from the residual that survives once plan and structure are accounted for, and it cannot be isolated from unobserved application quality, which is the channel it names. Political control is stated for the 2022–2026 data window; the May 2026 elections changed control in several of these boroughs. Where a borough's emerging plan or a single appeal would sharpen the verdict, the text says so rather than reaching past the record.