Housing demand in Canada: A novel approach to classifying mortgaged homebuyers

Context and motivation

House prices have grown rapidly in many parts of Canada in recent years. This has raised some concerns among policy-makers about financial stability and housing affordability. Yet, we know little about how different types of homebuyers have contributed to the dynamics of housing markets in Canada.

In this note, we document a novel approach to study home purchases in Canada. Combining different sources of microdata, we compute the share of mortgage-financed home purchases associated with three types of buyers:

  • first-time homebuyers
  • repeat homebuyers
  • investors

We then study the demographic and financial characteristics of these groups to derive initial insights into financial vulnerabilities associated with different types of homebuyers.

Data

We combine two sets of anonymized loan-level data, both available from 2014 onward:

  • Mortgage originations data: This dataset, obtained through regulatory filings of banks, contains the underwriting information of mortgages issued by a federally regulated financial institution.
  • Credit bureau data: This dataset, obtained from TransUnion, contains the credit histories of over 30 million Canadian residents.1

We develop a matching algorithm to detect the same borrowers across these two anonymized datasets. The algorithm includes variables such as the mortgage amount, the origination date, the lender’s name and the borrower’s forward sortation area (the first three characters in their postal code). We obtain a matching rate of close to 90% across most of the large banks.

To test whether the home purchases in our combined dataset are representative of home purchases in Canada, we compare our data with those of the Canadian Real Estate Association (CREA). As shown in Chart 1, the year-over-year growth rates of sales and prices in our dataset closely mirror their CREA counterparts—but with a slight lag. Such a lag is not surprising since mortgage funds are typically advanced when the property title is transferred, which happens after a sales agreement is finalized. Note, however, that purchases in our dataset are based on a different definition than those captured by the CREA. This is because we include mortgages issued for the purchase of new homes in addition to those for the purchase of existing homes. However, we do not capture purchases made using cash only or those made by corporations.

Chart 1: Matched data show growth dynamics similar to the usual resale statistics

Chart 1: Matched data show growth dynamics similar to the usual resale statistics

Sources: TransUnion, regulatory filings of Canadian banks, Canadian Real Estate Association and Bank of Canada calculationsLast observation: 2021Q2

Classifying homebuyers

With the matched dataset in hand, we then break down mortgaged home purchases into contributions from three distinct groups: first-time buyers, repeat buyers and investors.

First-time homebuyers

We identify purchases by first-time homebuyers by the first-ever appearance of a mortgage on a borrower’s credit file. We find that first-time homebuyers are the largest group of homebuyers, accounting for one-half of home purchases since 2014 (Chart 2).

Chart 2: First-time homebuyers generally account for about half of home purchases in Canada

Sources: TransUnion, regulatory filings of Canadian banks and Bank of Canada calculations

Repeat homebuyers

For a purchaser to be classified as a repeat homebuyer, the issuance of their new mortgage must also be associated with the discharge of a previous mortgage. Repeat homebuyers have accounted for 31% of home purchases since 2014.

Investors

This category represents homebuyers with multiple mortgaged properties. This means investors are homebuyers who either:

  • purchase an investment property while maintaining their primary residence, or
  • purchase a new residence to live in while converting their existing residence into an investment property2, 3

Since this dataset includes only mortgages originated by Canadian financial institutions, we capture mostly domestic buyers. Purchases from foreign buyers would be included only if the buyers obtained a mortgage in Canada. Under these parameters, investors have accounted for 19% of mortgaged home purchases since 2014.

Housing investments here may include the purchase of recreational properties, such as cottages. However, their inclusion does not alter our results in a material way. To investigate this question, we examine the share of investment purchases in non-urban regions.4 Specifically, we look at investors residing in 11 major cities in Canada (as in Chart 5) and find that their purchases in non-urban areas account for only 4% of all of their investment purchases since 2014. While small, this share has increased over time, from about 3% in 2014–15 to about 5.5% in 2020–21.

Homebuying patterns over time and across major cities

Next, we examine how home purchases from the three different groups have evolved over time. Chart 3 shows a high degree of co-movement in the growth rates of these groups’ home purchases. Interestingly, while purchases from all three groups have seen a rapid increase during the COVID‑19 pandemic, this is most pronounced for investors. The last time growth in the investor category outstripped that of first-time or repeat homebuyers was in 2017—during a period of exceptionally strong house price gains in Toronto and surrounding areas.

Chart 3: Home purchases by first-time homebuyers, repeat homebuyers and investors have historically moved in tandem

Sources: TransUnion, regulatory filings of Canadian banks and Bank of Canada calculationsLast observation: 2021Q2

As a result of these dynamics, the share of purchases by investors rose in 2017 and then again in 2021 (Chart 4).5 Currently, investors account for just over one-fifth of home purchases in Canada. Repeat homebuyers have also seen their share of activity increase slightly over time. In contrast, the share of purchases by first-time homebuyers has declined since 2015, reaching a new low in 2021; in the same six-year period, home prices have risen much faster than disposable income.

Chart 4: The share of home purchases by first-time homebuyers has declined since 2015

Sources: TransUnion, regulatory filings of Canadian banks and Bank of Canada calculationsLast observation: 2021Q2

Investor activity also varies across Canada’s major cities (Chart 5). At the low end, since 2014 investors have made 14% of home purchases in Winnipeg, compared with, at the high end, 21% in Toronto over the same period. We also note an upward trend over time in the share of home purchases by investors in nearly all major cities across Canada.

Chart 5: Investor activity has risen over time across major cities in Canada

Sources: TransUnion, regulatory filings of Canadian banks and Bank of Canada calculationsLast observation: 2021Q2

Homebuyer characteristics

To better understand vulnerabilities associated with different types of mortgaged homebuyers, we analyze their demographic and financial characteristics.

Starting with age, we see that first-time home buyers tend to be significantly younger than other types of homebuyers (Chart 6, panel a). Their median age is 36 years, compared with around 50 years for other types of homebuyers. The income distributions of the different types of homebuyer also reflect these life-cycle differences (Chart 6, panel b).

Turning to measures of financial vulnerability, we see that first-time homebuyers also tend to have the highest loan-to-income ratios at the time of mortgage origination (Chart 6, panel c). As a result, they greatly influence the total share of new mortgages with loan-to-income ratios above 450%—a key metric the Bank of Canada uses to monitor the vulnerability related to household indebtedness.

However, looking only at debt associated with the latest issued mortgage will tend to understate the financial vulnerability of investors who hold multiple mortgages. An advantage of our matched dataset is that we have the complete credit histories of all homebuyers. Therefore, we can observe the outstanding balances on all mortgages held by investors. When we recalculate loan-to-income ratios based on all mortgage debt, investors are clearly much more highly indebted than other types of homebuyers (Chart 6, panel d).

The larger debt load of investors is also likely reflected in the total debt service ratio calculated at the time the latest mortgage is issued. In principle, this measure should capture payments on all debt—including prior mortgages and non-mortgage debt.6 Chart 6, panel e, shows that investors tend to have higher total debt service ratios than non-investors. In particular, a noticeably higher share of investors have total debt service ratios above 44%.7 Investors that are highly indebted could face difficulty servicing their debt following a loss of income (either employment or rental) or an increase in interest rates.

An important gap in our analysis is that we cannot be certain what sources investors include in their documented income. Regulatory returns include only a single field for income, and it does not distinguish between employment income and rental income. Moreover, underwriting practices for rental income may vary across lenders. For example, some lenders allow applicants to use only 50% of rental income for mortgage qualification.

It is also unclear whether investors report all income when applying for a new mortgage or only enough to qualify. If income is underreported for investors with multiple investment properties, then panels d and e of Chart 6 may overstate the vulnerability. Without having more information on how rental income is being treated, we cannot easily assess the vulnerabilities and risks that investors bring to the housing market.

Chart 6: Housing investors tend to be older, earn more income and be more indebted

Chart 6: Housing investors tend to be older, earn more income and be more indebted

Distribution of homebuyers based on different metrics, 2014–2021H1

Sources: TransUnion, regulatory filings of Canadian banks and Bank of Canada calculations

Concluding remarks

In this note, we document the creation of a new dataset for tracking home purchases associated with a Canadian mortgage. The main advantage of this dataset is the ability to break down home purchases into the relative contributions of different types of homebuyers. A key insight to emerge from our initial analysis of this data is that home purchases are being driven increasingly by existing homeowners. Within this group, investors have seen the largest gain in their share of home purchases during the COVID‑19 pandemic.

The increased presence of investors in the housing market has contributed to strong demand and may reflect a belief that house prices will continue to rise in value—sometimes referred to as extrapolative expectations.8 Investors’ demand for housing may also be more sensitive to shifts in market sentiment than that of other homebuyers.9 By exacerbating so-called boom-bust cycles in housing markets, investors could thus be a source of instability for the financial system and the economy more broadly. At the same time, investors are an important source of housing rental supply. We need to do further research to examine the delicate balance between adding to rental supply while removing new builds and resale supply in a housing market that already has supply constraints.

More generally, the dataset has the potential to help answer many questions about household indebtedness and housing market imbalances, including the role of housing investors. This is an area of active research for the Bank.

  1. 1. To protect the privacy of Canadians, TransUnion did not provide any personal information to the Bank. The TransUnion dataset was anonymized, meaning it does not include information that identifies individual Canadians, such as names, social insurance numbers or addresses.[]
  2. 2. In this analysis, we have no information about the end use of the investment properties, which can be rented out on a long-term basis, rented out on a short-term basis through an online platform for vacation rentals, or left vacant.[]
  3. 3. Purchases by house flippers—individuals who buy and then resell homes a short time later—may also be captured here. However, according to Teranet data, homes bought and resold within six months have accounted for only approximately 1% of housing transactions at the national level in recent years. This share is around 2% for homes bought and resold within 12 months.[]
  4. 4. We define non-urban regions as areas located outside of census metropolitan areas or census agglomerates.[]
  5. 5. As a result of refinements in the algorithm, the series shown here differs slightly from the one in the 2021 Financial System Review.[]
  6. 6. The total debt service ratio compares required debt payments on all mortgages (principal and interest) and other products (such as installment loans, lines of credit and credit cards), as well as property taxes and heating costs, against the qualifying gross income of the borrower.[]
  7. 7. This 44% represents the total debt service cut-off for insured mortgages. For uninsured mortgages, total debt service limits are left to the discretion of individual lenders.[]
  8. 8. See U. Emenogu, C. Hommes and M. Khan, “Detecting Exuberance in House Prices across Canadian Cities,” Bank of Canada Staff Analytical Note No. 2021-9 (May 2021).[]
  9. 9. See for instance A. Haughwout, D. Lee, J. Tracy and W. van der Klaauw, “Real Estate Investors, the Leverage Cycle, and the Housing Market Crisis,” Federal Reserve Bank of New York Staff Report No. 514 (September 2011).[]

Disclaimer

Bank of Canada staff analytical notes are short articles that focus on topical issues relevant to the current economic and financial context, produced independently from the Bank’s Governing Council. This work may support or challenge prevailing policy orthodoxy. Therefore, the views expressed in this note are solely those of the authors and may differ from official Bank of Canada views. No responsibility for them should be attributed to the Bank.

JEL Code(s): R, R2, R21, R3, R31