The Backyard Is Being Built Over, Not Built In

Seven hundred and thirty permits, and what happened after

In 2018 the City of Vancouver issued 730 permits for laneway houses. The programme had spent a decade proving a modest idea: that a single-family lot could carry a second home if that home was small, in the back, and reached from the alley. By 2025 the city issued 365. On the 2026 run rate, measured through 22 September, it will issue about 245. That is a 66.4 per cent fall from the peak, and on its own it reads as a programme quietly expiring.

Read beside the permit next to it and the meaning inverts. Vancouver’s multiple-dwelling permits went from 114 in 2023 to 153 in 2024 to 376 in 2025 — the first year the multiplex out-permitted the laneway house, with 218 recorded by early October 2026 against 184 laneway permits. The backyard did not stop being built on. It stopped being the thing that was built.

That crossover is the finding of Pheonix Research’s new sizing of British Columbia’s small-scale multi-unit housing market, and it is the opposite of what the reform’s critics and its authors both expected. The province rezoned 1,147,785 single-family parcels to between one and six dwellings. The form that has answered is not the six-unit building the legislation pushed hardest for. It is the fourplex and the duplex, and the evidence for that is in a permit series that has been published, quietly, for eight years.

The same fixed cost, divided fewer ways

Every small multi-unit project on an infill lot carries a block of cost that does not move with the number of homes. Design, engineering and energy modelling run CAD 15,000 to 40,000. Permits, amenity charges and service connections run CAD 10,000 to 20,000. Servicing and site works, which on infill is frequently the largest single line because the old house’s water service was never sized for six kitchens, run CAD 15,000 to 50,000 before contingency. Add the interest carry across an approval period measured in months to years and the mid-point stack is roughly CAD 200,000 per site.

That block is the whole argument, because it is divided by door count and by nothing else. On a two-unit lot it is CAD 100,000 a door. On a six-unit lot it is CAD 33,333. Development cost charges then arrive per unit and refuse to behave any better: City of Vancouver residential rates reach CAD 80,000 to 100,000 a unit in higher-density zones, with Burnaby, Coquitlam and Surrey at CAD 40,000 to 70,000. Nothing else in this market triples on the same parcel.

Which is why the interesting question is not why so few six-unit buildings are being built. It is why the two-unit building is still winning when the arithmetic so plainly favours six.

What the permit values say that the permit counts do not

The laneway series carries a second number, and it is the one that explains the first. The median construction value declared on a Vancouver laneway permit rose 75.0 per cent in three years, from CAD 200,000 in 2023 to CAD 350,000 in 2026. Over the same window the product standardised hard: 97 per cent of recent permits are two-storey against 81 per cent in 2021 to 2023, and garage provision fell from 51 per cent to 27 per cent.

That is not a market raising its price. It is a market deleting its cheap end. The laneway house that survives into 2026 is a two-storey, two-bedroom, garage-free building costing CAD 350,000 — which is what it costs to build a proper small house on a serviced lot, and which is very close to the CAD 415,000 ticket the data shows on a duplex unit in a two-unit building. Once those two numbers converge, the choice stops being about money. It becomes a choice about whether the second home is behind the first one or beside it, and about whether the lot delivers two dwellings or four.

The convergence also marks the point where the laneway stopped being a budget product and became a design product. That is a real market, and it is not small. It is simply not the market the programme was invented to serve.

Why the front building wins

Three things decide it, and only one of them is construction cost.

The first is how the project is financed. A laneway house is almost always an addition: an owner-occupier with equity in a house, borrowing against it, no demolition, no new strata plan, no pre-sale requirement. A duplex is often still that — the owner lives in one half and sells or rents the other. A six-unit building is never that. It needs a development pro forma, a construction facility, an appraisal on completion value, and usually a strata plan. Those are different businesses, and the second one is where the volume is thin.

The second is that the six-unit building pays its fixed cost once and its approval risk six times. The servicing upgrade, the rezoning exposure, the neighbour appeal, the holding cost across a two-year approval — none of it is divided by anything, while the revenue it protects is spread across units that are, at the sizes this market builds, not especially expensive.

The third is municipality. Bill 44’s six-unit tier applies to parcels of 281 square metres or more within 400 metres of a prescribed frequent-transit stop, and it strips three municipal levers in exchange: no on-site parking requirement, density bonusing on one unit only, and no public hearing on the conforming bylaw. But the bylaw a given municipality actually wrote is the product a builder can actually get financed, and most of them wrote duplexes. The market has responded to the bylaw, not to the statute.

What the mix says about the law

Of the 6,400 units this report estimates were completed in British Columbia in 2025, the two-unit building accounts for 1,900 — 29.69 per cent of volume and 34.60 per cent of value, at CAD 415,000 a unit. The five-and-six-unit multiplex, the form the legislation was most aggressive about enabling, is 12.50 per cent of units and 14.04 per cent of value. The secondary suite, which needs no redevelopment at all, is 23.44 per cent of units for 12.51 per cent of value at CAD 190,000 a door.

Put differently: the reform’s most aggressive provision is the smallest product in the market, and the form most municipalities wrote by default is the largest. That is not a failure of the legislation. It is what happens when a province changes permission and leaves the delivery channel, the financing and the per-site cost stack exactly where they were.

There is also a geography to it, and it does not follow the map people assume. Metro Vancouver recorded 27,185 starts in 2025, down 3 per cent, and Victoria 4,859, up 16 per cent; Kelowna fell 31 per cent to 2,600 and Nanaimo 56 per cent to 448. The two southern coastal metros hold roughly seven starts in ten. But the reform’s marginal effect is largest where land is cheap relative to construction cost, because that is where a teardown can still pencil — the Okanagan, the Fraser Valley, and Vancouver Island outside the capital. It is smallest on Vancouver’s west side, where the land under a teardown is worth more than four units will recover, and where the permit response has been correspondingly thin. A municipality’s compliance with the legislation predicts almost nothing. Its land-to-build ratio predicts a great deal.

Four things that scale with the door, not the square metre

The reallocation matters beyond housing because the building that replaces a laneway house is not the same building with more units in it. It is a different envelope, a different glazing specification, a different bathroom count and a different fire strategy, and each of those has a market measured elsewhere.

Take the envelope first. A six-unit building has two or three exterior walls per unit against one or two in an apartment block, so the area that must be waterproofed, air-sealed and detailed rises faster than the floor area it encloses — which is what makes the ground-oriented form so much more waterproofing and construction chemicals per square metre than the tower it competes with for the same site. Glazing follows the same geometry: the window-to-floor ratio on a narrow three-storey form is high, and the energy step code sets the specification on top of that, so building and construction glass demand tracks this segment more closely than its floor area suggests. Bathrooms are simpler still: the count of suites, not the floor area, sets the order, which makes sanitary ware a per-door line. And four to six dwellings in one building need compartmentation and, in most configurations, sprinklers — a requirement no single-family house ever carried, and the reason fire protection enters the pro forma at exactly the point the door count passes two.

None of that shows up as a housing cost. All of it shows up in the material demand behind 10,833 units a year by 2033.

How the figures were built

The unit base comes from the one metropolitan area that publishes the split. Greater Victoria recorded 599 duplex and row starts out of 4,859 total in 2025 — 12.3 per cent of its starts — and that ratio, applied as a blended 10.5 per cent missing-middle share to the province’s 44,193 housing starts, produces the 6,400-unit base. Detached accessory units and secondary suites are added net of overlap with those starts, so no dwelling is counted twice.

The five forms are then sized individually and summed: 1,900 duplex units at CAD 415,000, 1,350 triplex and fourplex at CAD 385,000, 800 five-and-six-unit multiplex at CAD 400,000, 850 detached accessory units at CAD 430,000 and 1,500 secondary suites at CAD 190,000. Those shares tie back to the headline on both dimensions — 6,400 units exactly, and CAD 2,278.75 million of construction value, or CAD 356,055 a dwelling. Value per unit is escalated at 2.6 per cent a year to 2033, which is the gap between the 9.58 per cent value CAGR and the 6.80 per cent volume CAGR, and currency is held throughout at the Bank of Canada’s 23 October 2026 noon rate of 1.3993 Canadian dollars to the US dollar.

The resulting path of 69,624 dwellings from 2026 to 2033 is consistent with the province’s own modelling, which puts the combined SSMUH and transit-oriented development effect at 216,000 to 293,000 additional net-new units over ten years — the forecast’s delivery rate equates to an SSMUH share of roughly forty per cent of the low end of that range.

The direction is the part that does not depend on the arithmetic. The laneway house was the first product of a permission, and the permission has since been widened. The second product is bigger, it sits in front, and it is being built on the same lot with the same capital. The backyard is not being abandoned. It is being built over.