How the U.S., U.K., and Australia Measure Housing Starts Better Than We Do
How international definitions expose the weakness in Canada’s numbers
Highlights
Our housing start data is misleading us: Housing starts were up in 2025, but almost all the momentum was earlier in the year; sales have collapsed, and CMHC itself is warning that construction activity is now weakening.
Definitions are to blame: Canada counts a “housing start” far later in the construction process than peer countries, making starts a deeply lagging indicator, especially for high-rise projects.
In Canada, a building hasn’t “started” until it’s basically well underway: CMHC only counts starts once the foundation is complete and at grade, often a year or more after excavation begins.
Canada is out of step with our peers: The U.S., U.K., and Australia count starts much earlier, providing policymakers with a far more accurate, real-time picture of construction activity.
We need better data, not better caveats. CMHC should either modernize its definition of a housing start or publish a second, earlier-stage metric, because warnings in press releases are not substitutes for timely data.
When numbers diverge
Despite new housing sales all but collapsing in Canada’s largest markets, the CMHC released what appears to be a good news story: Housing starts up 5.6% in 2025 from 2024. The CMHC was careful, however, to note that Canadians should not draw the wrong conclusion, with their Chief Economist warning that starts data is about to deteroriate:
While housing starts in 2025 finished ahead of 2024 and inched up in December, most of the momentum in housing construction occurred in the Spring and Summer. Since September, the trend in housing starts has consistently decreased… housing starts are beginning this year from a weaker position and market intelligence suggests slowing momentum for residential construction.
This disconnect between sales data and start data is not simply due to the lag between preconstruction sales and project commencements; it reflects an underlying issue: Canada is out of step with our international peers in how we define a “housing start” and how the data is collected. Buildings and the units they contain are considered to have “started” much earlier in the United States, the United Kingdom, and Australia than in Canada.
Our peculiar definition of a housing start is leading us to misinterpret the state of the housing market, and the CMHC should make addressing this a priority.
In Canada, we start a year after we start
As we outlined in a previous piece, Canada’s definition of a “housing start” is not when construction begins on a new home. Instead, the Canadian Mortgage and Housing Corporation (CMHC), the agency responsible for collecting the data, does not count a housing project as a “housing start” until the foundation is complete and at grade. Once the building meets this definition, then every unit in that building is deemed being at the start of construction (so a 300 unit apartment building would be counted as 300 starts once the building meets the “foundation is complete and at grade” requirement).
Figure 1: Visual Definition of a Housing Start
Source: CMHC Housing Start Presentation to AMO.
For a high-rise project (either a purpose-built rental or condominium), it can take years, relative to when the developer obtains financing, for the building to reach the level of completion required to be considered a “housing start”. As such, it serves as a lagging indicator of the health of the housing market.
This is not the case in other countries.
Unlike in Canada, in the United States, a start measures the start of construction
In contrast with Canada, the U.S. Census Bureau and the U.S. Department of Housing and Urban Development do not require the foundation to reach grade before the units in a project are counted as a housing start. Instead, their definition only requires the start of excavation:
Start of construction occurs when excavation begins for the footings or foundation of a building. All housing units in a multifamily building are defined as being started when this excavation begins. Beginning with data for September 1992, estimates of housing starts include units in structures being totally rebuilt on an existing foundation.
In the construction of a large high-rise apartment with an underground parking garage, it can take up to a year from the start of excavation to when the foundation reaches grade. As such, the American definition of a start gives a much better real-time indication of the state of housing construction, particularly for high-rise projects.
The United Kingdom and Australia
The United Kingdom’s definition of a housing start is a hybrid of the U.S. and Canadian approaches. Unlike in the United States, excavation alone is not enough to count the units in a project as a start. But unlike Canada, in the United Kingdom, the laying of the foundation only just needs to have begun, rather than having to have fully reached grade:
A dwelling is counted as "started" on the date work begins on the laying of the foundation. It is counted as "completed" when it becomes ready for occupation or when a completion certificate is issued.
Australia’s version of a housing start is known as a “dwelling unit commenced,” whose measure more closely resembles that of the United States than Canada. Under the Australian definition, the start of excavation would be enough to qualify as a start, but unlike the United States, other activities can also cause the project to be added to the “dwelling unit commenced” data.
A building is commenced when the first physical building activity has been performed on site in the form of materials fixed in place and/or labour expended (this includes site preparation but excludes delivery of building materials, the drawing of plans and specifications and the construction of non-building infrastructures, such as roads).
Like Canada, the United States, and the United Kingdom, once a project meets this milestone, every unit in the building is deemed to have commenced.
Two potential paths forward
Governments and the media rely on housing start data to provide a near-real time metric of the state of new housing construction. Unfortunately, the way that Canada defines a housing start and collects this data creates considerable lag, which can mislead policymakers about the current health of the industry. Caveats, such as the quote in the recent CMHC release, are helpful but not enough.
Canada needs better real-time data on the current state of housing construction.
There are two paths forward for the CMHC in creating more timely housing start data, and each one is not without its challenges:
Option 1: Change the definition of “housing start” to one more aligned with that of the United States, the United Kingdom, or Australia. The downside to this approach is that it creates a discontinuity in the data, making housing start data before the change noncomparable to data after the change.
Option 2: Create a second metric and report data under both versions. For example, CMHC could collect “dwelling unit commenced” data, based on the Australian definition of the term. The upside of this approach is that it avoids the data discontinuity problem and provides us with additional insights into the lag between when a project commences and when a foundation reaches grade. The downsides are that it takes more work to collect two series, and it can cause confusion when two series attempt to measure the same thing.
Despite these challenges, it is absolutely vital for the CMHC to provide a more accurate, up-to-date picture of the state of new housing construction in Canada. Caveated quotes are not enough.