1. QUESTIONS

The location of different age groups has important implications for the social composition of cities. Studies of young adults observe a “youthification” effect as young adults are more likely to live in higher density neighbourhoods and move out to suburbs as they age keeping the higher density neighbourhoods “forever young” (Moos 2016; Lee 2022; Lee et al. 2024). Youthification intersects with gentrification and studentification but also operates as its own process (Moos et al. 2019; Revington et al. 2023). Youthification has been primarily studied in North America but also increasingly in different international contexts (Cocheci and Mitrea 2018; Shani and Bar-Haim 2026).

Although it is common to study residential geographies, no prior study has examined the age composition of different housing market segments explicitly. The association with young adults can reveal the ways in which certain housing market segments may act as “mechanisms of youthification” (Ma et al. 2018; Revington 2022). We ask:

How is the housing market segmented into different types of clusters?

What is the age composition of different housing market clusters?

We hypothesize that there is clear differentiation in the age composition among the different segments of the housing market with tall building neighbourhoods containing the highest share of young adults. We anticipate that findings will show the aging of the population in single-detached dwelling neighbourhoods and the youthification of tall building neighbourhoods.

2. METHODS

We use k-means cluster analysis to segment the housing market. A form of unsupervised machine learning, k-means cluster analysis categorizes data that are nearest to a defined centroid. The centroid is realigned and data re-categorized until the process reaches convergence (i.e., within-cluster variation is minimized). The number of clusters is defined by the user a priori. Four clusters are defined because they provide a good balance between minimizing within-cluster variation and producing meaningful and interpretable housing market segments.

We combine data spanning a 40-year period across Canada’s 20 largest census metropolitan areas (CMAs). The cluster analysis combines the 1981, 1991, 2001, 2011, 2016, and 2021 census data at the census tract (CT) level. This provides a combined data set of 30,678 observations (i.e., CTs) as a basis to determine the segmentation of the housing market. We use eight variables capturing tenure, dwelling type and dwelling size. For tenure we differentiate the percentage owned versus rented dwellings in a tract. Dwelling type is categorized as single-detached, single-attached, apartment in a duplex, apartment in buildings with fewer than 5 storeys, and apartment in buildings with five or more storeys. For dwelling size, we use the average number of rooms. This approach to housing market segmentation is unique for focusing on dwelling stock characteristics not price, rent, or turnover as is sometimes done.

Variables are converted to z-scores for the cluster analysis. The z-scores are generated by comparing an individual CT’s value on a specific variable to the average for that variable within the same CMA and census year. This is done to prevent creating clusters that are only found in one CMA or one census year. Finally, we identify the age composition of the different clusters in 2021 compared to 1981.

3. FINDINGS

As shown in Table 1, the first cluster, Low-Rise, is characterized by a higher share of rented apartments in buildings that have fewer than five storeys and a lower average number of rooms. The second cluster, Single-Detached, is characterized by owned single-detached dwellings with a higher average number of rooms . The third cluster, Tall Building, is characterized by rented apartments in buildings that have five or more storeys and a lower average number of rooms . The fourth cluster, Attached, is characterized as attached dwellings including townhouses, semi-detached houses, and duplexes as defined by the Canadian census. The single-detached dwelling is the most common dwelling type at 43 percent of all dwellings followed by apartments with fewer than 5 stories at 21 percent and apartments with five or more stories at 16 percent (Table 1).

Table 1.Dwelling Characteristics of Housing Market Clusters
20 Largest CMAs 2021 (percent) 2021 / 1981 (ratio)
Variable All CTs Low-Rise Single-Detached Tall Building Attached All CTs Low-Rise Single-Detached Tall Building Attached
% Owned 63 46 84 35 73 1.17 1.04 1.02 1.46 1.08
% Rented 37 54 16 65 27 0.80 0.97 0.89 0.86 0.84
% Single detached house 43 25 74 7 42 0.94 0.71 0.94 0.45 0.84
% Single attached 14 10 10 6 30 1.05 0.74 1.29 0.68 1.18
% Duplex 6 12 5 2 4 1.18 1.34 2.07 0.69 1.16
% Apartment, less than 5 storeys 21 42 8 18 17 1.01 1.30 1.25 0.80 1.35
% Apartment, 5 or more storeys 16 11 2 67 7 1.06 1.11 0.75 1.33 0.86
Average number of rooms per dwelling 6 5 7 4 6 1.06 0.98 1.09 0.89 1.01
0 Low - High 100 0.01 Low-High 2.00

Calculated using Statistics Canada census data.

The percent owning their dwelling is highest in the single-detached and attached clusters (84 and 73 percent respectively), and lowest in the low-rise and tall building apartment clusters (46 and 35 percent respectively). Our segmentation of the housing market shows the high degree of association between tall building neighbourhoods and renting, and single-detached dwelling neighbourhoods and owning.

Table 2 shows the age composition of the 20 largest CMAs, and Table 3 for the Toronto, Montreal, and Vancouver CMAs separately. The largest individual cohorts are the 25-34 and 35-44 year olds, followed closely by the 45-54 and 55-64 year olds. The smallest cohort are the 0-4 year olds.

The second to the fifth columns in Tables 2 and 3 show the age composition of each of the housing market clusters. In all CMAs combined, the low-rise apartment cluster has relatively higher shares of young adults and lower shares of teenagers. This is similar across the three largest CMAs. The single-detached cluster is most like the average across all CTs but shows some higher relative values of children and middle-aged and older adults. Lowest are the 25-34 year olds in the single-detached cluster.

The tall building cluster shows the most pronounced differences in age composition compared to other clusters. The tall building cluster has relatively higher shares of 20-24 year olds and 25-34 year olds and lower shares of children. Across all 20 CMAs, the shares of 45 to 69 year olds are lower than the CT average but the share 70 years and over slightly higher.

The largest increase over time has occurred among the older cohorts with the aging of the population. Shares of older cohorts increased in the single-detached and attached clusters. The tall building cluster was the only cluster to see an increase in young adults 25–34 years of age between 1981 and 2021, although it was only a slight increase in magnitude. The cohorts aged 35 to 44 and 70 and older also increased within the tall building cluster.

There is a clear age structure visible in the housing market with the tall building segment most directly defined by young adults. There is a strong link between tenure and housing types, as single-detached dwellings are linked to ownership and taller building clusters to renting. We find evidence of youthification as young adults continue to be associated with the higher density dwelling types.

Table 2.Age Composition of Housing Market Clusters, 20 Largest CMAs
20 Largest CMAs 2021 (percent) 2021 / 1981 (ratio)
Variable All CTs Low-Rise Single-Detached Tall Building Attached All CTs Low-Rise Single-Detached Tall Building Attached
% Aged 0-4 years 5 5 5 4 5 0.74 0.82 0.67 0.84 0.66
% Aged 5-9 years 6 5 6 4 6 0.83 0.88 0.77 0.86 0.74
% Aged 10-14 years 6 5 7 3 6 0.78 0.74 0.77 0.72 0.74
% Aged 15-19 years 6 5 6 4 6 0.61 0.53 0.65 0.54 0.64
% Aged 20-24 years 6 7 6 8 6 0.63 0.59 0.72 0.66 0.67
% Aged 25-34 years 14 17 11 22 13 0.79 0.95 0.66 1.08 0.69
% Aged 35-44 years 14 14 13 15 14 1.09 1.27 0.91 1.31 1.01
% Aged 45-54 years 13 12 14 11 13 1.20 1.10 1.25 1.05 1.32
% Aged 55-64 years 13 13 15 11 13 1.48 1.28 1.73 1.09 1.85
% Aged 65-69 years 5 5 6 5 5 1.59 1.34 2.04 1.07 2.24
% Aged 70 years and over 12 12 12 13 11 1.95 1.71 2.56 1.31 3.06
0 Low - High 100 0.01 Low-High 4.00

Calculated using Statistics Canada census data.

Table 3.Age Composition of Housing Market Clusters, 3 Largest CMAs
Toronto CMA 2021 (percent) 2021 / 1981 (ratio)
Variable All CTs Low-Rise Single-Detached Tall Building Attached All CTs Low-Rise Single-Detached Tall Building Attached
% Aged 0-4 years 5 5 5 4 5 0.72 0.74 0.74 0.75 0.62
% Aged 5-9 years 5 5 6 4 6 0.79 0.79 0.81 0.80 0.68
% Aged 10-14 years 6 5 6 4 6 0.76 0.71 0.79 0.69 0.72
% Aged 15-19 years 6 5 7 4 6 0.65 0.57 0.67 0.56 0.70
% Aged 20-24 years 7 7 7 7 7 0.71 0.64 0.80 0.64 0.81
% Aged 25-34 years 15 16 12 23 13 0.83 0.88 0.78 1.08 0.71
% Aged 35-44 years 14 14 13 16 14 1.04 1.17 0.90 1.26 0.93
% Aged 45-54 years 14 13 15 12 14 1.18 1.14 1.13 1.07 1.35
% Aged 55-64 years 13 13 14 11 13 1.46 1.37 1.47 1.18 1.90
% Aged 65-69 years 5 5 5 4 5 1.55 1.45 1.70 1.19 2.15
% Aged 70 years and over 11 12 11 11 11 1.89 1.86 2.06 1.46 2.72
Montreal CMA 2021 (percent) 2021 / 1981 (ratio)
Variable All CTs Low-Rise Single-Detached Tall Building Attached All CTs Low-Rise Single-Detached Tall Building Attached
% Aged 0-4 years 5 5 5 4 5 0.79 0.93 0.61 1.11 0.74
% Aged 5-9 years 6 5 6 4 6 0.92 1.02 0.75 1.21 0.86
% Aged 10-14 years 6 5 7 3 6 0.83 0.79 0.79 0.91 0.78
% Aged 15-19 years 5 4 6 4 6 0.59 0.49 0.64 0.70 0.58
% Aged 20-24 years 6 7 5 9 5 0.60 0.57 0.67 0.89 0.59
% Aged 25-34 years 14 17 10 20 11 0.77 0.99 0.57 1.00 0.66
% Aged 35-44 years 14 15 14 14 13 1.05 1.26 0.85 1.17 0.91
% Aged 45-54 years 13 12 14 10 13 1.12 0.99 1.34 0.91 1.15
% Aged 55-64 years 13 12 15 11 14 1.44 1.14 2.24 0.88 1.64
% Aged 65-69 years 5 5 6 5 6 1.58 1.26 2.76 0.87 1.93
% Aged 70 years and over 13 12 12 16 15 2.19 1.86 3.82 1.26 2.93
Vancouver CMA 2021 (percent) 2021 / 1981 (ratio)
Variable All CTs Low-Rise Single-Detached Tall Building Attached All CTs Low-Rise Single-Detached Tall Building Attached
% Aged 0-4 years 4 4 4 4 5 0.71 0.70 0.63 1.34 0.71
% Aged 5-9 years 5 4 5 3 6 0.78 0.79 0.73 1.23 0.81
% Aged 10-14 years 5 4 6 2 6 0.73 0.72 0.72 0.92 0.76
% Aged 15-19 years 5 5 6 3 6 0.64 0.60 0.69 0.61 0.64
% Aged 20-24 years 7 7 6 7 6 0.70 0.66 0.85 0.55 0.67
% Aged 25-34 years 16 18 11 25 14 0.85 0.93 0.68 1.09 0.71
% Aged 35-44 years 14 15 12 17 15 1.10 1.32 0.82 1.52 1.08
% Aged 45-54 years 13 13 14 12 14 1.25 1.27 1.26 1.13 1.32
% Aged 55-64 years 13 13 15 12 13 1.38 1.28 1.61 0.97 1.51
% Aged 65-69 years 5 5 6 5 5 1.36 1.17 1.77 0.83 1.67
% Aged 70 years and over 12 12 13 12 11 1.60 1.33 2.47 0.84 2.12
0 Low - High 100 0.01 Low-High 4.00

Calculated using Statistics Canada census data. Toronto n=1,222. Montreal n=986. Vancouver n=529.

ACKNOWLEDGMENTS

We would like to thank the participants at the American Association of Geographers conference in Detroit for comments on earlier drafts of this work. We acknowledge the financial support from the Canada Social Sciences and Humanities Research Council grant on high-rises led by Ute Lehrer at York University. We also thank the peer-reviewers for their helpful comments and careful review.