Canada's Housing Crisis Isn't About Starting Homes. It's About Finishing Them.

77 years of data reveal the real bottleneck isn’t starts, it’s the backlog of units taking forever to build.

Photo by Mario Verduzco on Unsplash

TLDR

• Units under construction average 40,271/month, roughly 7x higher than starts or completions.
• All three metrics correlate strongly (r=0.73, 0.80), confirming a predictable but cyclical pipeline.
• November 2021 saw the all-time peak of 25,014 housing starts in one month.

The hidden scale of Canada’s construction backlog

174,929 units. That’s the single-month peak for housing under construction, hit in October 1973. But here’s what makes that number quietly horrifying: it’s not the exception. It’s the normal state of Canada’s building system.

For nearly eight decades, the Canadian housing pipeline has been dominated by one metric that almost nobody talks about: the massive inventory of units stuck mid-construction. While starts and completions average around 5,800 units per month each, units under construction sit at roughly 40,271. You read that right. Seven times bigger.

This isn’t a bug in the data. It’s the architecture of Canadian housing itself.

Canada Housing Trends Over Time (Centres 10,000+): Starts, Under Construction & Completions

The data spanning 1948 to 2025 shows that Canada’s housing system isn’t broken because we’re not starting enough homes. We’re stuck because the ones we start take forever to finish. The pipeline is clogged, not empty.

Seven times more units are under construction than we start each month

Why the numbers correlate, but the timelines don’t

You’d expect a chaotic system to produce random numbers. Instead, the data shows something stranger: housing starts, units under construction, and completions all move together. The correlation between completions and units under construction hits r=0.802. The rest hover between 0.73 and 0.80.

But here’s the catch. Strong correlation doesn’t mean fast pipeline. It means predictable slowdown.

Correlation Matrix: Housing Starts, Under Construction & Completions (Total Units)

Strongest Correlation: Housing completions vs Housing under construction (r=0.802)

When you start more homes, you do eventually complete more. But the lag time, the duration those units spend under construction, is what’s actually gotten longer over time. The correlation is tight. The timeline is stretched.

So when policymakers point to November 2021’s record 25,014 starts as proof the system is working, they’re reading only half the equation. Yes, we’re starting. But what happened to the 2020 backlog surge units? Many are still under construction.

The 1970s boom: when Canada built 174,929 units in one month

The all-time peak wasn’t last year. It was October 1973, during an inflationary spiral and a housing craze that feels almost alien by today’s standards. That single month saw nearly 175,000 units under construction simultaneously across Canada’s centres with 10,000+ people.

Canada Housing Under Construction — Anomalies (Z-Score > 2.5)

Three months total flagged as statistical anomalies across the entire 77-year dataset, and the 1970s boom is the most extreme. Construction was chaotic then too, but in the opposite direction: lots happening at once, few guardrails, prices spiraling. The system maxed out.

Then it crashed. Hard. By the 1990s, Canada’s housing construction entered a prolonged slowdown that felt like a different country altogether. And after 2020, a new surge. The pipeline fills, empties, fills again. But the pattern is always the same: steady starts, then a long wait for completions.

Canada Housing Starts (Total Units) — Outliers Highlighted

Apartments dominate the numbers, but they hide the real story

If you look at total value by housing type, apartments and multi-unit dwellings completely dwarf everything else. They’re the volume play. Semi-detached units, row houses, and specialized housing types barely register on the graph.

Total Units by Housing Estimate Type and Unit Type

Canada Housing: Total Values by Housing Estimate Type and Unit Type

Top 10 Segments by Total Value

But volume isn’t the same as accessibility. Apartments take longer to construct than single-family homes. Multi-unit buildings require more inspections, more coordination, more time sitting idle waiting for permits or materials or labor. The units under construction metric gets pulled upward by apartment complexity.

This matters because policy focused on starts per se misses why completions lag. You can fast-track a bungalow subdivision. You cannot fast-track a 20-story condo tower with the same speed.

Apartments dominate starts but maximize construction delays

The 2021 peak, the 2020 backlog, and where we are now

November 2021 marked the highest single-month housing starts on record: 25,014 units across all of Canada’s mid-to-large centres. That was real demand. New policies, low rates, immigration, pent-up need. All converging.

Canada Housing Starts (Total Units) — Annual Trend 1948–2025

But look at what didn’t happen next. Those 25,014 starts didn’t immediately convert to 25,014 completions in 2023 or 2024. Instead, they joined the backlog. The units under construction metric surged in the post-2020 period and stayed high because starts outpaced completions. More stuff started. The queue got longer.

The three structurally significant macro-patterns the data reveals are clear. The 1970s boom. The 1990s bust. And the post-2020 backlog surge we’re living in right now. This one’s different because it’s sustained by policy intervention and demographic pressure, not a speculative bubble about to pop.

What the distribution really tells us

The data is wildly uneven. Standard deviation of 22,263 units against a mean of just 12,749. Nearly twice as high. Some months, construction barely happens. Other months, it explodes. Some regions surge while others stall. Some housing types get investment; others vanish.

Distribution of Housing Values by Estimate Type and Unit Type

Housing Statistics Summary

This variability isn’t noise. It’s structural. It reflects regional inequality, cyclical boom-bust patterns, and the uneven way capital and labor flow through Canadian housing markets. Toronto and Vancouver move at different speeds than Atlantic Canada or the Prairies. Apartment construction clusters; single-family home starts scatter.

The histogram of housing values shows the distribution is heavily right-skewed. Most months are quiet. A few months are insane. The same pattern you see in stock prices or income inequality, but for housing. Power-law territory.

Top 10: Highest Total Values by Housing Estimate & Unit Type

Bottom 10: Lowest Total Values (>0) by Housing Estimate & Unit Type

The real bottleneck isn’t starting homes. It’s finishing them.

Here’s what the 77-year record actually shows: Canada’s housing crisis isn’t primarily a problem of insufficient starts. It’s a problem of duration. Projects that take too long. Pipelines that clog. Permits that drag. Labor and materials that stall.

The November 2021 record proves we can start homes fast when demand is high and capital is cheap. The months after 2021 prove that starting and completing are not the same thing.

The solution isn’t just “build more.” It’s “finish faster.” Streamline permitting. Unblock labor. Stabilize materials costs. Because every month a unit sits under construction, it’s not serving anyone. It’s capital locked in limbo. It’s housing that exists on paper but not in neighborhoods.

The correlations in this data are tight. But the timelines are loose. That’s the real problem Canada faces.

Want to dive deeper? Explore the full interactive data story on Datastam.

[embed]Canada Mortgage and Housing Corporation, housing starts, under construction and completions in… Canada's housing dataset spanning 77 years (1948-2025) reveals a strongly right-skewed construction landscape dominated…canada.datastam.ai

Originally published on Datastam.