The Least Affordable Ten

The Affordable Ten ran a natural experiment: hold price roughly constant at the bottom of the market and watch whether outcomes still fan out. They did, by forty points, which is how that essay knew price could not be the mechanism. This one runs the experiment again at the opposite end. The ten most expensive boroughs in London sit in a band that starts above the richest of the affordable ten and climbs to more than a million pounds, and the question is whether, once a borough's housing is this dear, the outcome is settled. The companion paper found that approval tracks price across London; if that gradient is a law, the boroughs up here should all say yes.

Almost all of them do. Nine of the ten approve above what their caseload predicts, several of them comfortably, which is the wealth mechanism the first essay identified working at full strength. But the experiment's value is in its single failure. One of the ten most expensive boroughs in London sits nine points below expectation, and it is the one that has written its restriction into an adopted plan. The gradient is real, but it is a tendency, not a law, and the borough that breaks it shows exactly what overrides it.

§1 Why these ten: holding price high

These are the ten boroughs with the highest median house price in London, running from Barnet and Southwark in the mid-£540,000s up to Kensington & Chelsea at £1.1 million. The band is deliberately the inverse of the affordable ten's: where those ten sat inside a £107,000 window at the bottom, this group occupies the entire top half of London's price distribution, and its cheapest member is dearer than the most expensive of the affordable ten. If affordability drove outcomes mechanically, this would be the most uniformly permissive group in the city. It is mostly, but not quite, that, and the "not quite" is the essay.

Nine green bars and one red. The nine are the wealth mechanism the first essay set out: boroughs whose well-resourced applicants clear demanding design and heritage tests as a matter of course, so that even plans which read as neutral or restrictive produce generous outcomes. The tenth, Barnet, receives the same kind of well-resourced application and refuses a measurable share of it anyway. The rest of this essay is about why.

§2 The mirror experiment

It helps to set the two price experiments side by side. The affordable ten held price in the low £400,000s and found outcomes spanning roughly forty points, from Havering at −21 to Bexley at +19, proof that at the bottom of the market, affordability cannot be what decides. This group holds price above £545,000 and finds outcomes spanning thirty points, from Barnet at −8.7 to Richmond at +21.8. The range has narrowed (wealth has lifted the whole group and removed the deep negatives that dominate the affordable end) but it has not closed. A thirty-point spread at constant high price is still a spread that price cannot explain, and the same residual the affordable ten chased at the bottom is present, smaller but unmistakable, at the top.

What has changed between the two ends is the balance of the categories, not their existence. At the bottom, restriction was common and permissiveness the exception worth explaining. At the top, permissiveness is the rule and restriction the exception. But in both cases the exception is the informative case, because the exception is the borough doing something its price does not predict, and in both cases, as it turns out, the borough doing it has a plan that explains the move.

§3 What the number hides: a sea of yes, one row of no

Disaggregated to site type, the group is overwhelmingly green. Conversions, rebuild, backland, mid-terrace: across the expensive boroughs the cells sit above expectation, the broad and uniform generosity the first essay traced to applicant quality rather than to any single permissive rule. Two rows interrupt the pattern, and both are informative. Camden runs a deliberate split (generous on conversions, negative on demolish-and-rebuild) because its plan welcomes subdivision and subjects demolition to a retention test. And Barnet is a red row almost on its own: below expectation on conversions and rebuild alike, the only borough in the group whose dominant colour is the wrong one.

Barnet's signature is the conversion signature from the previous essay, transplanted into the wealthiest tier of the city: hard on conversions, ordinary on rebuild, with a named family-housing policy behind it. The fact that the same fingerprint appears among the affordable boroughs and again at £545,000 is the point. The mechanism that produces it (a deliberate refusal to let family houses become flats) is available to a borough at any price, and Barnet is the borough rich enough to be expected to wave such schemes through that has instead chosen to gate them.

§4 The price gradient and its one exception

Plotting the whole city makes the gradient and its exception visible at once. Across London, approval and price move together (the cloud slopes up, the correlation is real if loose) and the expensive boroughs cluster, as expected, in the generous upper-right. Barnet is the point that has fallen out of the cluster: as expensive as Wandsworth and Hackney, and as restrictive as boroughs worth half as much. It is the mirror image of Bexley, the cheap borough that floats above the line because its plan says yes. Barnet sinks below the line because its plan says no.

§5 Barnet: the rich borough that says no

Barnet is among the most expensive boroughs in London and one of its largest, a borough of interwar suburbs whose applicants are as well-advised as anyone's, and it sits nine points below expectation on small sites. The plan audit explains it without residue. Barnet's adopted Local Plan carries Policy HOU03, which permits the conversion of a three-to-five-bedroom family house into flats only where a family-sized unit of at least seventy-four square metres is re-provided at ground floor, a direct brake on the standard house-to-flats scheme. Behind it sits a borough-wide Article 4 direction, in force since 2016, that removed the permitted-development route to small houses in multiple occupation, so that conversions which elsewhere proceed without an application must in Barnet be applied for and judged against the policy. And over the top of both lies a long-standing suburban-character framework, anchored in the borough's characterisation study, that drags demolish-and-rebuild down as well. The conversion column is the reddest because the conversion policy is the sharpest; the rebuild column is mildly negative because the character framework is softer. The data is the policy made visible.

Two further facts make Barnet the decisive case rather than a curiosity. It is not restricting because it is failing to build: it passes the Housing Delivery Test at 104 per cent, so the family-housing brake is a choice made from delivery comfort, not a symptom of a stalled pipeline. And its restriction is, unlike the unexplained boroughs of the second essay, genuinely written down, a borough that has decided wealth is not a reason to surrender its family housing, and has said so in an adopted plan it can defend at examination. Barnet is the rich refuser done legitimately. It is what it looks like when a borough uses the one lever its price does not control.

§6 The appeal record: rare refusals, marginal refusals

The appeal record behaves as the trilogy's general finding predicts. The permissive boroughs in this group lose a larger share of the few appeals they face (Richmond is overturned at better than a third of decided cases, Kensington & Chelsea at a comparable rate on the broader appeal set) not because their decisions are poor but because a borough that approves most of what it receives refuses only the schemes it is least sure of, and those marginal refusals are the ones an Inspector is most likely to allow. Barnet, by contrast, is overturned at a little over a quarter of decided appeals, in line with the city rather than above it: its refusals are not the rare marginal ones of a permissive borough but the routine application of a standing policy, and they hold up accordingly. As everywhere in this series, the appeal rate describes the texture of a borough's decisions, not their legitimacy; the reason Barnet says no is in its plan, and the reason the others say yes is in their applicants, and the Inspectorate confirms both without adjudicating either.

§7 The spine: Richmond versus Barnet

The Affordable Ten's spine was Bexley against Havering: two cheap boroughs, forty points apart, one saying yes on the page and the other saying no off it. This essay's spine is the same contrast carried to the top of the market. Richmond and Barnet are both expensive, both receiving the well-prepared applications that wealth brings, and they land thirty points apart. Richmond's fresh plan adopts the small-sites duty and its wealthy applicants clear its character gates, so it says yes twice over, by plan and by purse. Barnet's plan protects family housing and it enforces the protection regardless of who is asking, so it says no on the one ground a borough fully controls. Neither outcome is set by price, because their prices are nearly the same. Both are set by the plan.

§8 What this means: the plan is what defies the price

Set the five essays' corner cases on a single grid and the whole series resolves into one claim, one we can now state simply. Price runs along one axis: at the bottom, Havering and Bexley; at the top, Barnet and Richmond. Outcome runs along the other. Three of the four corners go where price predicts: Havering is cheap and restrictive, Richmond is dear and permissive, and the great mass of boroughs in between follows the gentle upward slope. But two corners defy their price, and they are the two that matter. Bexley is cheap and permissive because an adopted policy encourages small homes. Barnet is dear and restrictive because an adopted policy protects family ones. The boroughs that escape the price gradient are precisely the boroughs that have written a small-sites policy and meant it, in opposite directions, from opposite ends of the market, by the same mechanism.

That is the practical residue of the whole exercise. Price is a strong prior and a weak rule: it tells you where a borough will probably land and never where it must. The thing that lets a borough sit somewhere its price does not predict (Bexley above its station, Barnet below it) is a plan that takes a clear position on small sites and is enforced regardless of who is applying. Wealth is not in a council's gift and neither is poverty, but the plan is, which is why the two boroughs that defied their prices did it the same way, and why the lever worth attending to is never the price tag but the policy.

§9 Conclusion

The affordable ten asked whether outcomes vary when price is held low, and found that they vary enormously. The least affordable ten ask whether they vary when price is held high, and the answer is the same, only quieter: wealth lifts the whole group and removes the deep refusals, but a thirty-point spread survives, and at its foot sits a borough as expensive as its permissive neighbours and as restrictive as boroughs worth half as much. Barnet does not break the price gradient by accident. It breaks it on purpose, through an adopted policy that protects family housing from a market that would otherwise convert it away, and it does so while meeting its housing targets and holding its refusals at appeal. The gradient found the tendency; Barnet found its limit. Across all five essays the same instrument keeps returning the same lesson: that the borough number is where a question begins, and that the answer, at every price and every outcome, is written not in what a borough's homes cost but in what its plan decides to do about them.

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); the price–differential correlation across the 31 boroughs with comparable medians is Pearson r ≈ +0.53. Appeal overturn: LSSPD matched-appeal cohort (London 21.0%; Richmond 36.7%, Barnet 27.3%). Housing Delivery Test: Barnet 104% (Pass, 2023 measurement). Plan audit: Barnet Local Plan (adopted 4 March 2025, Policy HOU03; borough-wide HMO Article 4 in force 2016; Characterisation Study 2010); the nine permissive boroughs as audited in The Permissive Ten and The Middle Ten (Richmond, Kensington & Chelsea, Hammersmith & Fulham, Camden, Southwark, Haringey from the former; Islington, Wandsworth, Hackney from the latter).

What this can and cannot tell you. The descriptive claims are firm: these ten boroughs have the highest median prices in London, nine approve above their caseload-expected rate and one (Barnet) approves well below it, and Barnet's restriction maps onto a named, adopted conversion policy plus a borough-wide Article 4 direction. The "wealth lifts the group" reading is the inference developed in The Permissive Ten (that better-resourced applications clear demanding plans) and it cannot be isolated from unobserved application quality, which is the channel it names. The price–outcome correlation is moderate (r ≈ +0.53), so price is a tendency rather than a determinant; Barnet is the worked demonstration of the residual, not a refutation of the gradient. Political control and plan status are stated for the 2022–2026 data window; several boroughs in this group changed control in May 2026, after the last decision in the data. The figures locate where a borough sits relative to its price; they do not predict the outcome of any individual application.