No 3-Bedroom Homes, No Kids, No Future: Why Families Are Leaving Cities
The data is clear - families with kids want to own homes that have three bedrooms or more
Highlights
Housing policy, particularly in Ontario, is plagued by “unit-is-a-unit” thinking, which treats a studio condo as equivalent to a 3-bedroom home.
However, homes with a minimum of three bedrooms are necessary for many families with kids to meet Canada’s “adequate space” standards in their implementation of housing as a human right.
Examining data for Southern Ontario in the 2016 and 2021 Censuses, we observe that increases in any form of housing are positively correlated with population growth. However, only the growth of ownership-based homes with three or more bedrooms is strongly correlated with the population growth of children under the age of five.
There is a negative correlation between the growth in 3+ bedroom rental housing and the growth of the population of children under the age of five, as a “conversion to rental” of existing homes crowded out young families.
The causal relationship between the growth of 3+ bedroom owner-occupied housing and the growth in the population of children is two-sided. Population growth leads to increased demand for housing, which is met by the market. Causality also works in the opposite direction, as communities that create the conditions for the growth of family-sized homes attract young families, and also allow for existing families to have more children.
The tax, regulatory, and economic conditions in many Ontario communities are making it increasingly difficult for middle-class families to afford a three-bedroom home, resulting in a decline in the population of young children in those communities. This needs to be fixed.
Families seek out family-sized homes that they can own
Housing type matters.
A point we have made repeatedly at The Missing Middle, in pieces such as "The Baby Bust" and "The Death of the Three-Bedroom Ownership Home," is that in order to attract and retain families with young children, communities need to allow for the construction of homes, for ownership, with three or more bedrooms.
Our conjecture about the relationship between the growth of 3+ bedroom homes and the growth of young families can be tested with data. Using Census data, let’s examine the strength of the correlation between the two in Southern Ontario.
At the Missing Middle, we’re both fans of maps and the Census, so whenever we have an opportunity to combine the two, we’ll take it. Between Census 2016 and Census 2021, there were substantial shifts in Southern Ontario in the population of children under the age of five. As shown in Figure 1, the population of young children decreased in the GTA and Ottawa, but increased dramatically in areas such as Lanark Census Division and Oxford County.
Figure 1: Change in Population of 0-4 Year Olds, Census 2021 vs. Census 2016, by Southern Ontario Census Division
Source: Author’s calculation, from Census 2016 and 2021 data
It’s important to note here that anyone who was aged 0-4 in the 2016 Census would have aged out of that category by the 2021 Census, so we are measuring two entirely different cohorts of children. Which leaves us with the question, “Why did some places see the size of the next cohort of children rise, while others experienced a decline?”
We conjecture that housing availability played a role. Let’s test that.
Correlating babies to housing
Using data from the 2016 and 2021 censuses, we calculated population growth rates and housing stock growth rates for Southern Ontario’s 38 census divisions. Specifically, we examined the following measures:
The percentage change in the population of 0-4 year olds
The percentage change in the population of 5+ year olds
The percentage change in the number of 0-2 bedroom homes, broken down by owner-occupied and rental properties.
The percentage change in the number of homes with three or more bedrooms, broken down by owner-occupied and rental properties.
In Figure 2, we show the metrics for four Census Divisions: Ottawa, which experienced a small decline in the number of children under age five (but a large increase in older populations), Hamilton, which experienced a small growth in the number of young kids, and Lanark (near Ottawa) and Oxford (near Hamilton) which experienced high growth rates in the number of children.
Figure 2: Changes in Population and Housing Stock, Census 2021 vs. Census 2016, by Southern Ontario Census Division. (Values above 10% in green, values below 3% in red)
Source: Author’s calculation, from Census 2016 and 2021 data
For measuring growth in housing, we are not using housing starts; rather, the Census measures the growth in the number of occupied homes, which, unlike housing starts, accounts for demolitions and other changes in the housing stock. The Census measures whether owners or renters occupy the home, not whether it is a purpose-built rental, which allows us to capture changes to how homes are used. The big increase in three-bedroom rentals in Ottawa and Hamilton is almost (but not completely) entirely due to formerly owner-occupied homes being converted into rental properties. Census Divisions with high growth in student populations, such as Waterloo Census Division and Middlesex (London) Census Division, experienced growth rates in 3+ bedroom rental homes exceeding 20%.
Correlation coefficients between our 11 measures are shown in Figure 3. The correlation between the growth in the population of 0-4 year-olds and those over the age of four is 0.52. We should not be surprised that places experiencing population growth among people above the age of four also added young children; the correlation between the two is relatively modest. In other words, the number of young kids a place added is not simply a function of that place’s overall population growth.
Figure 3: Correlation Coefficients Between Changes in Population and Housing Stock, Census 2021 vs. Census 2016, for Southern Ontario’s 38 Census Divisions (Values above 0.6 in green, negative values in red)
Source: Author’s calculation, from Census 2016 and 2021 data
We highlighted particularly strong correlations in green and negative correlations in red to help identify patterns more easily. There were some surprising results, including a very small but negative correlation between increases in owner-occupied homes and rental homes; that is, there was a slight inverse relationship between the number of owner-occupied homes and the number of renter-occupied homes in a Census Division.
Examining the correlation matrix, five relationships jumped out to us as particularly important.
What our baby-housing correlations showed
There was almost no relationship between the number of 0-2-bedroom homes created in a community and changes to the population of children under five. Communities that built a high number of these units did not increase (or decrease) their population of young children compared to communities that built few.
There was, however, a strong correlation between increases in the housing stock and increases in the population of those over the age of four. Building any form of housing was positively correlated with population growth, and the correlation between population growth (among individuals aged five and above) and the growth of the total housing stock was extremely high at 0.9.
The metric most highly correlated with the population growth of children under the age of five was the increase in the stock of owner-occupied housing with three or more bedrooms, with a correlation coefficient of 0.68.
Changes in the stock of 3+ bedroom owner-occupied housing and 3+ bedroom rental housing were negatively correlated. This is because much of the increase in the stock of these homes resulted from investors purchasing them and renting them out, primarily to international students and other populations with a high propensity to rent.
There was also a negative correlation between 3+ bedroom rental housing and the growth of the population of children under the age of five, as this “conversion to rental” crowded out young families who traditionally would have purchased these homes.
This is the point in the article where someone states, “correlation is not causation.”
And they would be right. Correlation is not causation. There is also a fairly obvious potential causal relationship: communities with a high number of young families, who are reaching the age where they are starting to have children, tend to build a large number of family-friendly homes, as developers respond to this market demand. That is absolutely part of the story.
We believe the causality also works in the opposite direction: communities that build a lot of family-friendly housing attract families with young children. To test this, we obtained intraprovincial (within-province) migration figures for children aged 0-4 years, between 2016 and 2021, from Statistics Canada's Table 17-10-0153-01. During these five years, the City of Toronto experienced a net intraprovincial outmigration of 29,233 children under five, while Peel Region experienced a net outmigration of 8,450 children, and Durham Region experienced a net in-migration of 6,440.
We normalized these population flows for the size of the 0-4 population as of 2016 and created a new metric, “Intra 0-4”, which measures the intraprovincial migration of children between the ages of 0 and 4 from 2016 to 2021. Not surprisingly, we find that the migration of these children is strongly correlated (r = 0.71) with the growth of the overall population of children aged 0-4 years. We also find a strong positive correlation (r = 0.59) between the migration of young children and the stock of 3+ bedroom owner-occupied homes, and a weak but negative correlation between the growth of 3+ bedroom rental stock and the migration of young kids, further evidence of the displacement effect.
Figure 4: Correlation Coefficients Between Changes in Population and Housing Stock, Census 2021 vs. Census 2016, Adding Intraprovincial Migration of 0-4 Year Olds, for Southern Ontario’s 38 Census Divisions (Values above 0.6 in green, negative values in red)
Source: Author’s calculation, from Census 2016 and 2021 data
And, yes, we do recognize the irony of adding another correlation analysis to a section that includes the phrase “correlation is not causation”.
There is a third potential causal relationship, that access to larger housing causes families already in the community to have more children than in communities where access to larger homes is limited. We did not test this conjecture; however, some academic studies have examined it, including a study that compared fertility rates to housing typologies in Finland.
Overall, though, the evidence for Southern Ontario is quite clear. “Drive until you qualify” is real, and young families with kids (or wanting to have kids) move to communities where they can access homes with three or more bedrooms.
This isn’t simply an economic issue; it’s a human rights one
As the piece Housing as a Human Right Requires 3+ Bedroom Homes in Every Community made clear, Canada’s National Housing Strategy Act recognizes adequate housing as a human right under international law. For families with multiple children, “adequate housing” typically requires a home with three or more bedrooms. Families are migrating out of cities such as Toronto, because that right cannot be realized.
We should acknowledge this as the crisis that it is, and policymakers at all levels of government must eliminate the barriers that prevent the construction of family-sized homes in our communities.
Download a PDF of this article here:
