Economic and Social Reports
Renters’ shelter costs by duration of tenancy

by Search for Statistics Canada publications by Samuel MacIsaac, Samuel MacIsaac , ORCID for Samuel MacIsaac and David Wavrock

DOI: https://doi.org/10.25318/36280001202500800003-eng

Skip to text

Text begins

Abstract

Renters’ shelter costs can vary substantially by duration of tenancy. Renters who recently moved into their dwelling can pay substantially more than longstanding renters. This tenure-based discrepancy in rents paid can have important implications in terms of discouraging residential mobility and, by extension, Canadian labour mobility. Using data from the Census of Population and the National Household Survey, this study sheds light on trends in shelter cost differences by duration of tenancy from 1996 to 2021, and how this gap in rental costs between recent renters and longer-term renters varies across municipalities and neighbourhoods. Findings show that the gap between recent and long-term renters’ shelter costs has widened over the study period and that tenure-based gaps in rental costs are pervasive beyond the priciest housing markets of Toronto and Vancouver. Accounting for comparable dwellings and neighbourhoods, longstanding tenants of five or more years paid approximately 19% less in monthly shelter costs than recent renters (under one year) in 2021 compared with 6% in 1996. The percentage of renters residing in the same dwelling they did five years ago increased from 30% to 42% of all renters during that period.

Keywords: Rent, housing costs, affordability, tenant turnover, price stickiness

Authors

Samuel MacIsaac and David Wavrock are with the Economic and Social Analysis and Modelling Division, Analytical Studies and Modelling Branch, at Statistics Canada.

Acknowledgments

The authors would like to thank Aled ab Iorwerth at the Canada Mortgage and Housing Corporation and Jean-Philippe Deschamps-Laporte and several researchers from the Social Analysis and Modelling Division at Statistics Canada for their comments on previous versions of this study.

Introduction

Approximately one in three Canadians rent their primary residence. Rental costs can vary significantly, with new renters facing substantially higher rents than longstanding renters (Gorski 2023; MacIsaac 2023; Canada Mortgage and Housing Corporation [CMHC] 2024a; Statistics Canada 2024). This partially reflects the contrast between rapidly rising rents for new rentals and capped rent increases for existing renters, whose rent may be subject to controls,^Note  thereby resulting in rents typically increasing more during tenant turnover (Gorski 2023). Housing rent rigidity, or stickiness in housing prices among long-term tenants compared with recent renters, can also be the result of landlords seeking to minimize the risk of having a vacant unit (Gallin and Verbrugge 2019), avoid the cost of finding a new tenant or retain a “good tenant,” among other possible motives.

The discrepancy between relatively rigid rents for long-term tenants and substantially higher market rates for recent tenants could have important implications in terms of affordability and residential mobility. For instance, tenants who move—whether they do so willingly or because they are forced to—often face steep housing cost increases, which can impose substantial added costs for renters. Moreover, substantial rent increases during tenant turnover may outpace potential affordability gains related to rental sorting mechanisms, whereby the cost of rental units decreases with the age of the dwelling and consequently increases the supply of affordable units for lower-income families (CMHC 2024b).

Potentially large rent increases could dissuade longstanding tenants from moving, and this may hinder labour mobility. Gellatly et al. (2024) show that rising housing prices have disproportionately affected young Canadians’ moving decisions, particularly those of renters. Evidence from Sweden suggests that high relative housing prices contributed to lower labour mobility in Stockholm and that the effects were more pronounced among young and highly skilled workers (Eliasson and Westerlund 2024). This could inhibit efficient labour matching, whereby skilled individuals are optimally matched with job vacancies across Canada, and this could have knock-on effects on productivity.

Gaining a better understanding of how the gap between prices for recent and long-term tenants has changed over time and across Canadian municipalities is a priority within this context. Using data from the Census of Population and the National Household Survey (NHS), this study provides an overview of the costs borne by renters in Canada, by duration of tenancy, by addressing the following two research questions:

1. How have renters’ shelter costs changed by duration of tenancy from 1996 to 2021?
2. How does the gap in rental costs between recent renters and longer-term renters vary across municipalities?

This article is divided into four sections. The first section outlines the data sources, derived variables and limitations of the analytical sample. The second presents tabular results and trends in tenure-based rental pricing gaps from 1996 to 2021, complemented by a third section that refines the estimation of this gap using multivariate techniques to account for confounding factors. The fourth section provides concluding remarks and implications for future research.

Data

This study employs data from the Census of Population (1996, 2001, 2006, 2016 and 2021) and the 2011 NHS, with rented primary residences as the main unit of analysis. The sample is restricted to private dwellings with renters who paid non-zero rents and that were geographically located within the 41 census metropolitan areas (CMAs) in 2021.^Note  The sample also excludes dwellings built during the previous five calendar years or during the census year because these new dwellings were not available to longstanding tenants of five years or more but are common across recent renters, and this could hinder comparability.^Note  For 2021, this yields a sample of approximately 815,000 unweighted rental dwellings.

Several sources report on rental market rates, which capture the cost of newly listed properties available for rent. Market rates offer timely insights into the costs borne by prospective renters, but do not account for the typically lower rental costs of longstanding tenants. To address this difference, a variable for duration of tenancy is derived from two census questions asking whether respondents lived at the same residence that they did a year prior to the given census reference date and whether they lived at the same address that they did five years prior to the reference date. This variable indicates whether a respondent has resided in their current dwelling for less than one year, for one year to less than five years, or for five years or more.^Note  A tenancy of less than one-year acts as a proxy of market rates, which can then be compared with rents paid by longer-term tenants.

The census provides valuable information on shelter costs and the duration of tenancy for rented dwellings from 1996 to 2021. The data also include numerous dwelling characteristics that act as control variables, and the large sample facilitates disaggregation at a more granular geographic level. For comparability across rental units, the main variable of interest is shelter cost, which includes both rent and utilities, to account for variability in the inclusion of electricity, heating, water and other municipal services as part of rent.^Note  These data complement those from CMHC’s Rental Market Survey, which provide valuable information about the gap in average rents between turnover (last 12 months) and non-turnover units (CMHC 2024a).

However, there are some limitations to these data and hence the interpretation of this study’s findings. Reliance on data from specific census years introduces temporal constraints, especially given that the analysis ends in May 2021, when there were numerous rental market disruptions caused by the COVID-19 pandemic, such as a relatively high number of vacancies^Note  (CMHC 2024a) and temporary protections for renters in some provinces.^Note 

Other data limitations involve omitted variables. For instance, known factors tied to variability in rental costs, such as the rental property ownership structure (Teresa 2019; St-Hilaire et al. 2024), are unavailable when using census data. Moreover, the primary (i.e., purpose-built rentals) and secondary rental markets cannot be analyzed separately when using census data. Another limitation is that data on whether a dwelling is subsidized are available only from 2011 onward. Lastly, the duration of the tenancy variable remains relatively crude for those who resided at the same address for more than five years—key distinctions between tenure intervals exceeding five years (e.g., 5 to 10 years, 11years or more) are overlooked.^Note 

Shelter cost gaps between recent and longer-term tenants

From 1996 to 2021, average shelter costs in Canadian census metropolitan areas rose substantially. For dwellings with a tenancy of less than a year, average shelter costs rose from $640 per month in 1996 to $1,480 per month in 2021, while those with tenures of five years or more saw shelter costs increase from $600 per month in 1996 to $1,100 per month in 2021 (Chart 1). Rental turnover declined alongside this growth in the gap between shelter costs by duration of tenancy. This marks a shift toward a larger proportion of tenants living at the same address for five years or more. The share of renters who lived at a different address a year earlier dropped from 29.5% to 19.9% of renters (Chart 2).

This carries important implications for Canadians’ living arrangements and their residential mobility. The rising gap in shelter costs between recent and longstanding renters and the shift from 29.5% to 41.9% of renters residing in the same dwelling they lived in five years earlier from 1996 to 2021 point to growing groups of renters that are susceptible to large increases in their housing costs (e.g., new renters facing rapidly rising rents).

Table 1 further disaggregates the results from Chart 1 for five census divisions (CDs): Toronto, Montréal, Greater Vancouver, Calgary (Alberta, Division No. 6) and Ottawa.^Note  In 2021, the average shelter costs for recent renters (i.e., less than a year of tenancy) were 34% higher than for renters with tenancies dating back five years or more across Canadian census metropolitan areas. By comparison, the same gap was 52% in Toronto, 41% in Ottawa, 39% in Vancouver, 31% in Montréal and 10% in Calgary. In 1996, the gap for Canadian census metropolitan areas was around 7%, and there was no significant difference in shelter costs between the two groups in Montréal. There is a similar tenure-based gap in shelter costs between rented dwellings with tenancies of less than one year and those with tenancies of one year to less than five years, but it is expectedly smaller.

While descriptive statistics provide valuable insights into average shelter costs overall, they could mask a variety of compositional effects across dwellings. For instance, thus far, the analysis has not discounted differences in shelter costs by duration of tenancy potentially reflecting recent renters living in pricier, recently built dwellings.^Note  Another possible explanation could be that recent renters have inherently different dwelling and neighbourhood characteristics, compared with long-term renters. For example, of the 41% of all rented dwellings in the bottom 20% of neighbourhoods by median income in 2021, less than one-fifth housed tenants with a tenancy of less than one year, compared with over two-fifths that housed tenants with a tenancy of five years or more. To address this, the next section provides a more in-depth look at the relationship between duration of tenancy and shelter costs using multivariate hedonic pricing models that account for renters’ dwelling and neighbourhood characteristics.

Shelter costs by duration of tenancy accounting for compositional effects

To better understand how shelter costs vary in accordance with differences in dwelling and neighbourhood characteristics, Table 2 presents multivariate analysis results using hedonic pricing models. Hedonic pricing models estimate the value of a certain good (e.g., monthly shelter costs as the dependent variable) as determined by its constituent characteristics (e.g., type of dwelling, number of bedrooms) and the sum of their contributory values (Keshishbanoosy and Taylor 2019; MacIsaac 2023). Ordinary least squares regressions are used to estimate five models. Models 1 to 3 use shelter costs in 2021 as the dependent variable, while models 4 and 5 use the natural logarithm of shelter costs in 2021 and 2016, respectively, as dependent variables.^Note  Models were estimated using survey weights and cluster-robust standard errors to account for unobserved similarities within each dissemination area (DA).

Model 1 is a model without any controls and shows that, without control variables, dwellings with a tenancy of five years or more cost $373 less per month on average than dwellings rented out in the last year. However, this overstates the gap attributable to duration of tenancy, which requires comparing dwellings with similar characteristics. The addition of control variables for these characteristics in models 2 and 3 reduces this difference to between $237 to $266 per month. Though smaller, this gap remains substantial and represents an annualized difference of $2,844 to $3,192. The gap between dwellings rented within the last year and those with tenancies of one year to less than five years was $73 to $85 per month, corresponding to an annualized difference of $876 to $1,020 (models 2 and 3).

Differences between models 2 and 3 reflect the choice of neighbourhood control variables. While Model 2 includes neighbourhood (i.e., DA level) median income quintiles and accounts for census subdivision fixed effects, Model 3 includes DA fixed effects. The effects on coefficients between the two models are similar, but goodness of fit (i.e., the R-squared value) is somewhat better in Model 3. This likely reflects controlling for more granular neighbourhood characteristics beyond differences in median income. Note that these fixed effects likely capture various local regulatory differences (e.g., zoning laws in the case of DA-level controls) and market differences (e.g., vacancy rates) across geographies, which act as valuable controls for differences in rents paid.

Other control variables include typical dwelling characteristics such as the number of bedrooms, dwelling type, dwelling condition and whether the dwelling is subsidized. Like in other comparable studies (Keshishbanoosy and Taylor 2019; MacIsaac 2023), dwellings with fewer bedrooms, those with longer tenure, attached housing, dwellings that require more maintenance and dwellings in lower-income neighbourhoods cost less, all other factors being equal. Although not shown in Table 2, the inclusion of neighbourhood controls had the largest effect in reducing the coefficient for the tenure-based gap in monthly shelter costs between recent renters and those with a tenancy of five years or more.

In terms of comparisons across years, models 4 and 5 use the natural logarithm of shelter costs as the dependent variable, yielding coefficients for duration of tenancy that are more easily comparable across years. Without using logarithmic transformations, coefficients expressed in dollar values would not account for inflationary differences across years. In 2016, the coefficient for a tenancy of five years or more was -0.096, which corresponds to longer-term renters paying 9.1% less than recent renters when accounting for comparable dwellings and neighbourhoods (Model 5). In 2021, the results indicate that longstanding tenants of five years or more were paying around 16.1% less in monthly shelter costs, compared with recent renters (Model 4). This marks a notable increase in the gap from 2016 to 2021, with cities such as Greater Vancouver, Toronto, Montréal and Ottawa experiencing higher tenure-based gaps (see Appendix Table A1). There are only minor differences in the coefficients of control variables across both models.

Separate models were run to study this gap starting in 1996. This analysis was run separately from results for 2016 and 2021 in Table 2—earlier years required a pared-down model because the 1996, 2001 and 2006 censuses did not collect information on whether renters’ dwellings were subsidized. The 2011 NHS was excluded from all multivariate analyses because it differs substantially in its sampling. The results are presented in Appendix Table A2, which provides comparable estimates from 1996 to 2021, despite these limitations.^Note  Though the tenure-based gap is smaller when accounting for confounding factors, Table A2 exhibits the same trend as in Chart 1: the gap widened from 1996 to 2021. Moreover, a separate dynamic decomposition model was created to determine the portion of the shelter cost gap between longstanding and recent tenants from 1996 to 2021 that is attributable to changes in prices and the portion that is attributable to changes in rental dwelling composition. This decomposition model of the difference in shelter costs from 1996 to 2021 confirms that the growth in the gap was indeed attributable to changes in shelter costs rather than differences in composition (i.e., location or dwelling characteristics). The latter compositional effects accounted for less than a tenth of the gap.^Note 

The tenure-based gap in shelter costs is a pervasive issue across Canadian municipalities. Though monthly shelter costs in the Toronto and Vancouver rental markets were markedly higher than in many other Canadian cities in 2021, the tenure duration-based gap in percentage terms was similar for many other cities. For comparable dwellings and neighbourhoods in 2021, renters residing in a dwelling for five years or more in Montréal and Ottawa paid around 18% less than recent renters who had resided in their home for less than a year (see appendix). This was similar to the 20% difference in Toronto and Vancouver.

Conclusion and discussion

Rental price rigidities in Canada favour longstanding tenants who pay substantially less rent than more recent tenants. This study shows how the gap between recent and long-term renters’ housing costs has widened since 1996. Moreover, the continued fast-paced increase in rental costs since 2021 suggests that this gap may have widened further in recent years. Since May 2021, many of the temporary provincial rent control policies and tenant protections enacted following the onset of the pandemic have come to an end. Wide differences in rents paid between turnover apartments (new tenant in last 12 months) and non-turnover apartments for 2023 appear to corroborate this trend (CMHC 2024a).

This study finds that, accounting for dwelling and geographic fixed effects (e.g., differences in vacancy rates, neighbourhoods, regulations), tenancy duration remains an important factor for tenants’ rent costs. The implication is that renters may increasingly be incentivized to avoid moving, and this may hinder residential and labour mobility. Another implication could be dissuading otherwise qualified individuals from applying to employment opportunities that require frequent relocations, such as the military. This lower labour mobility could have knock-on effects on the ability to efficiently allocate labour and skills across Canada and, therefore, labour productivity more broadly. Looking ahead, further research is needed to determine the effects of such rental cost gaps on renters’ residential mobility and tenant turnover.

Appendix

References

CMHC (Canada Mortgage and Housing Corporation). (March 25, 2020). COVID-19: Eviction bans and suspensions to support renters. CMHC.

CMHC (Canada Mortgage and Housing Corporation). (2024a). Rental Market Report. CMHC.

CMHC (Canada Mortgage and Housing Corporation). (2024b). Understanding Filtering: A Long-Term Strategy to New Supply and Housing Affordability. CMHC.

Eliasson, K., and Westerlund, O. (2024). Housing markets and geographical labour mobility to high-productivity regions: The case of Stockholm. European Urban and Regional Studies, 31(3), 259-280.

Gallin, J., and Verbrugge, R. J. (2019). A theory of sticky rents: Search and bargaining with incomplete information. Journal of Economic Theory, 183, 478-519.

Gellatly, G., Foran, H., and Pinault, L. (2024). Barriers to moving: Potential implications for the life satisfaction of young families. Economic and Social Reports, 4(12), 1-5.

Gorski, A. (2023). A tale of two renters: Housing affordability among recent and existing renters in Canada. Statistics Canada.

Government of Nova Scotia. (September 6, 2024). Changes to rent cap, Residential Tenancies Act.

Keshishbanoosy, R., and Taylor, L. (2019). New approach for estimating the rent component of the Consumer Price Index. Statistics Canada.

MacIsaac, S. (2023). What’s included in Canadians’ rent? Economic and Social Reports, 3(8), 1-17.

Statistics Canada. (2024). Housing affordability in Canada, 2022. Statistics Canada.

St-Hilaire, C., Brunila, M., and Wachsmuth, D. (2024). High rises and housing stress: A spatial big data analysis of rental housing financialization. Journal of the American Planning Association, 90(1), 129-143.

Teresa, B. F. (2019). New dynamics of rent gap formation in New York City rent-regulated housing: Privatization, financialization, and uneven development. Urban Geography, 40(10), 1399-1421.