CMHC April Starts Jump 17% but New Permit Data Shows Pipeline Still Subdued

The 17% month-over-month spike is real, and it is large. April's seasonally adjusted annual rate of 279,317 units is a meaningful jump from March. But CMHC itself flags why the monthly SAAR can mislead in isolation: the multi-unit segment, which now dominates Canadian construction, swings sharply from month to month as large condominium and rental projects break ground. A single high-rise can move the national number.

That is why CMHC pairs the monthly figure with a six-month trend, which smooths the volatility. The April trend reading rose to 256,777 units, up 3.2% from March. That is the more reliable read on momentum. It follows several months of decline. Improvement, yes. Breakout, no.

The non-annualized actual starts in centres of 10,000 or more tell a similar story. April 2026 came in at 21,805 units — about 1% below April 2025. Year-to-date is up 6% versus the same period last year, but that growth is concentrated earlier in the year and in specific provinces. The April spike in the headline number is not a sudden national surge; it is a multi-unit-heavy month in a few large markets, layered onto a still-fragile underlying trend.

A single national percentage can hide three different markets. April made that obvious. Toronto's actual housing starts rose 34% year-over-year. Montréal climbed 21%. Vancouver fell 30%. Multi-unit activity — condominiums and purpose-built rentals — drove most of the difference in all three cities.

For homeowners watching their own market, the national headline is the wrong reference point. The metro number is the real signal, and within the metro number, the mix of project types matters: high-rise condominium and purpose-built rental projects move on different timelines and respond to different financing conditions than detached construction.

The structural change in this release is the new permit indicator. CMHC is now publishing — for centres with populations of 50,000 or more — the number of dwelling units with approved building permits where construction has not yet begun. These approved-but-not-started units are dwellings with permits issued (excluding conversions) where no work has commenced, and they are not yet counted as housing starts. They are, in CMHC's framing, a forward-looking measure of supply.

In April, that pipeline shrank. Across all large centres in Canada, the count of approved-but-not-started units fell from 148,839 in March to 142,324 in April — a 4.4% drop in a single month. Ontario, Canada's largest housing market, saw an even sharper contraction: from 33,592 approved-but-not-started units to 29,408, a 12.5% monthly decline. At the same time, the stock of units actively under construction in Ontario edged up modestly.

Read those two movements together and the picture sharpens. The under-construction stock is high because projects approved in prior years are still working their way through. But the queue behind them — the pipeline of permits waiting to convert into starts — is thinning. Today's building activity reflects yesterday's decisions. Tomorrow's activity will reflect today's pipeline, and that pipeline is shrinking.

CMHC has been making the lead-lag case explicitly. Chief Economist Mathieu Laberge has explained that building permits serve as an early signal of future residential construction activity, while housing starts capture projects that have actually begun and therefore offer the best read on near-term supply. Releasing the two together, he argues, gives a more complete view of the construction pipeline than either does alone.

Deputy Chief Economist Kevin Hughes was direct about what the April pipeline data shows. While the six-month trend in starts ticked up slightly, it followed several months of decline. The new permit data, he said, points to a construction pipeline that remains subdued, in line with the agency's most recent Housing Market Outlook.

The lag matters in concrete terms. CMHC's analysis notes that high-rise condominium and purpose-built rental projects typically see roughly nine to fifteen months between permit issuance and actual start. Single-detached homes move faster — closer to two to ten months. Those lags vary by city. What they mean is that a drop in approved-but-not-started units today shows up in the starts data roughly one to two years from now, not next month.

There is also a compositional reason permits have become a sharper indicator. Before about 2012, single-detached construction dominated Canadian starts. From 2012 to 2025, condominiums took over as the largest category. Since 2025, purpose-built rental apartments have led — longer-lead projects with more decision points before ground breaks. The slower a project moves through its life cycle, the more useful early-stage data like permits becomes.

The Spring 2026 Housing Supply Report from CMHC makes a related point worth holding alongside the April release. Short-term conditions in some markets look almost relaxed — record completions, higher vacancies, slower rent growth. Those conditions reflect strong starts from earlier years now reaching completion. The future pipeline, by the same report, is thinning. Starts have declined for two consecutive years on the broader trend. Project cancellations, particularly in the condominium segment, will reduce the number of units delivered several years out.

CMHC has been blunt about why. Land scarcity, high construction and regulatory costs, and weak pre-construction sales have pushed the financial viability of many projects underwater. Some developers are cancelling. Others are delaying or shifting to less expensive locations. The structural reasons the permit pipeline is shrinking are not mysterious — they show up in every viability calculation a developer runs.

That structural backdrop produces a tension the brief frames explicitly. A thinning pipeline supports values in supply-constrained markets, because fewer new units competing on the market means less downward pressure on existing prices. The same thinning pipeline worsens conditions for buyers and renters trying to enter, because the additional units that would soften affordability are not arriving. Both effects can be true at once, in the same city, in the same quarter. That is the dynamic April's data captures.

The April release is two stories layered on each other. The first is the 17% bounce, which will dominate coverage and which is real for the month it happened. The second is the new permit indicator, which says the pipeline feeding the next two years of construction got smaller, not bigger, in the same month. The first story is what is being built right now. The second is what will not be built later if the trend holds.

For existing homeowners, the practical read is interpretive, not predictive. Densification pressure and the supply-side dilution of property values are functions of future starts, not current ones. A shrinking permit pipeline implies fewer of those future starts. Whether that translates into measurable price effects, neighbourhood change, or affordability outcomes depends on demand conditions that are not in this dataset. CMHC's own Housing Market Outlook projects continued decline in new home construction through 2028 as builders face higher costs, weaker demand, and rising inventories of unsold homes, particularly in the condominium segment. April's permit data is consistent with that forecast, not against it.

It is also worth holding the bigger frame in view. CMHC's supply gap analysis estimates that returning to 2019-level housing affordability would require between 430,000 and 480,000 new units built annually for a decade — roughly double the current pace. The agency's earlier shortage research puts the cumulative gap at about 3.5 million additional units needed by 2030, on top of what is already projected to be built. Against that benchmark, even a sustained 279,317-unit annual pace would leave the affordability shortfall growing. A pipeline that is thinning makes the gap wider, not narrower.

The April 2026 data is one month. The trend is up. The pipeline is down. The signal that matters is not which way the headline ran in the past 30 days; it is what the permit data implies for the next 24 months. CMHC's new disclosure puts that question directly on the table. The answer it suggests is not the answer the headline number gave.