# Analyzing the house price boom in the suburbs of Canada’s major cities during the pandemic

## Introduction

“Location, location, location,” says the old real estate adage.

House prices tend to be higher in downtown areas of major cities and gradually decline as one moves farther away from the city centre. This is known as the bid rent theory, first introduced by American economist William Alonso (1964). This theory is rooted in the fact that:

• land is scarce in downtown areas
• the potential for high foot traffic implies that retailers are willing to bid big money to secure a lease

Land closer to downtown therefore tends to be more expensive. But in addition to these theoretical considerations of land use, downtown living is often characterized by:

• shorter commutes
• easy access to public transit and services1

These advantages may also contribute to the so-called proximity premium, which typically associates higher prices with houses located near downtown.

But this pattern may have shifted during the COVID‑19 pandemic. The move to remote work in many sectors of the economy made living close to the office less advantageous. Many of the services that downtown residents typically enjoy—such as restaurants, salons and gyms—were closed at various times during the pandemic. And between working or studying from home and the public health restrictions, people were spending more time at home than ever before. A desire for more living space may have encouraged many Canadians to seek properties in the suburbs, where lots and houses are typically larger and more affordable.

To evaluate how much the cost of residential properties has changed during the pandemic based on their location, we compare the price of houses sold in Canada based on their distance to the closest city centre, controlling for various neighbourhood characteristics. In that respect, we are not trying to explain overall growth in house prices during the pandemic. Instead, we assess changes in relative prices between different locations. We find the following:

• As the bid rent theory predicts, house prices tend to be lower the greater the distance from downtown.
• Before the pandemic, the price gap between houses in the suburbs and those downtown was closing slowly but steadily.
• This house price gap narrowed faster during the pandemic, consistent with a preference shift toward more living space.
• If the preference shift is temporary, house prices in the suburbs could face downward pressure. This could be especially true if the construction of new houses in these areas were to greatly increase in anticipation of local demand continuing to rise.

## Data

To analyze the relationship between the price of houses and how close they are to downtown, we use land registry data compiled by Teranet and National Bank.2 For each forward sortation area (FSA) in Canada, they calculate the average price of houses sold in a given quarter.3, 4 These data include transactions for all types of dwellings—whether they are single- or multiple-family homes.

To locate these FSAs in relation to city centres, we calculate the average distance between each FSA and the closest city centre among Canada’s 15 major census metropolitan areas (CMAs):5

• St. John’s
• Halifax
• Moncton
• Saint John
• Québec
• Montréal
• Ottawa
• Toronto
• Winnipeg
• Regina
• Edmonton
• Calgary
• Vancouver
• Victoria

During the pandemic, house prices increased strongly in most neighbourhoods. But that growth was stronger in the suburbs (Chart 1 and Figure 1).

### Chart 1: House prices have increased more in suburbs than downtown

Sources: Teranet, National Bank and Bank of Canada calculationsLast observation: 2021

Figure 1: Growth in house prices in the Toronto and Montréal regions has been stronger in suburbs than downtown

Growth in house prices between 2019 and 2021 (%) for all property types, by forward sortation area

a. Toronto region

b. Montréal region

Sources: Teranet, National Bank and Bank of Canada calculationsLast observation: 2021

## Estimation

To evaluate how much the proximity premium has changed over time, we estimate the following equation separately for each quarter between 2014 and 2021:6

$$\displaystyle{\log\,(price}_{i,t})$$ $$\displaystyle=\,\alpha_{j,t}CMA_i$$ $$\displaystyle+\,\beta_t\log{\left(1+distance_i\right)}$$ $$\displaystyle+\,\gamma_tX_i$$ $$\displaystyle+\,e_{i,t}$$ $$\displaystyle,$$

where:

• $$i$$ captures each FSA (995 in total)
• $$j$$ captures each major CMA (15 in total)
• $$t$$ represents each quarter between 2014 and 2021
• $$price_{i,t}$$ is the average price of houses sold in the $$i$$th FSA in each quarter
• $$CMA_i$$ is a dummy variable for the major CMA closest to the $$i$$th FSA7
• $$distance_i$$ is the distance of the $$i$$th FSA from the city centre of the closest CMA (in kilometres)
• $$X_i$$ captures control variables for the $$i$$th FSA
• $$e_{i,t}$$ is the error term for the $$i$$th FSA in each quarter

Because we are specifically interested in measuring how the distance from the downtown core affects house prices, we restrict our exercise to the FSAs located within 80 kilometres of the centre of the major CMAs in our study.8

The control variables included in this exercise differ across FSAs but reflect their value in the 2016 Census of Population. They capture:9

• a measure of after-tax income10
• the average age of the population
• the share of immigrants in the population
• the share of dwellings represented by condominiums
• the share of dwellings represented by:
• apartments in a building that has five or more storeys
• single detached houses
• the share of homeowners with a mortgage

The measure of interest in this study is $$\beta_t\,$$, which is often referred to as the bid rent gradient. It represents the marginal impact of distance from downtown on house prices and is expected to be negative. This implies that the price of residential properties tends to decline the further the house is from downtown.

## Results

Chart 2 shows that the estimated value of the bid rent gradient is negative, as expected, but the absolute value is declining throughout the period. This suggests that the distance-related price premium of city centres has weakened over time.11 However, this narrowing in the price gap appears to have gained additional momentum during the pandemic.

### Chart 2: The closing of the price gap between houses in the suburbs and downtown gained momentum during the pandemic

Source: Teranet, National Bank and Bank of Canada calculationsLast observation: 2021Q4

Chart 3 shows the average estimated reduction in house prices for each kilometre relative to living downtown (kilometre zero). For instance, in 2016, houses in a neighbourhood with similar characteristics but located 50 kilometres from downtown would typically sell for 33% less than those located directly in the downtown core. In 2019, this price reduction had moved to 26%. By 2021, if the trend in pre-pandemic narrowing of the price gap had continued, the reduction in prices would have been 21%; however, given the developments during the pandemic the estimated reduction was only about 10%.

### Chart 3: The premium associated with living close to downtown fell considerably during the pandemic

Sources: Teranet, National Bank and Bank of Canada calculationsLast observation: 2021

## Conclusion

We analyze how location affects the market value of houses in Canada. We find that, as the bid rent theory predicts, houses located in the suburbs of major cities tend to have a lower price than houses in neighbourhoods with similar characteristics but located downtown. During the pandemic, this price gap shrunk considerably. This faster narrowing of the price gap based on geographical location reflects a shift in preferences during the pandemic toward more living space. It may also reflect the necessity for some households, especially among first-time homebuyers, to move farther from downtown to afford a home in a time of record-high house prices.

Better understanding why this price gap is shrinking faster could help shed light on the risks that the housing market and the economy face as the pandemic fades. On the one hand, if the increase in housing demand during the pandemic in suburban areas reflects a lasting preference shift, the reduction observed in the proximity premium could become permanent. On the other hand, if this preference shift is temporary, the proximity premium could return partly toward its pre-pandemic level. Such a shift in relative prices could be especially problematic if housing supply in more suburban areas were to respond strongly in anticipation of local demand continuing to increase.12

1. 1. In a survey conducted by the City of Toronto in 2011 among nearly 34,000 households living downtown and in the suburbs (Scarborough, North York, Etobicoke and Yonge-Eglinton), the top four reasons for choosing to live downtown were “Close to work,” “Access to public transit,” “Access to shops / stores / market,” and “Ability to walk everywhere.”[]
2. 2. As a robustness check, we also consider the price registered by financial institutions when they issued a new mortgage. Estimation results and findings were similar.[]
3. 3. Teranet and National Bank use these prices to construct their house price index.[]
4. 4. Forward sortation areas are geographical areas identified by the first three characters of their postal code. The 2016 Census of Population included 1,620 FSAs.[]
5. 5. FSAs are represented geographically by the centre point of the area, whereas the downtown core of each CMA is represented by the location of its city hall. We calculate the distance in kilometres as the straight-line distance (rather than driving distance).[]
6. 6. This exercise is similar to the one performed by Gupta et al. (2021) for major US cities. These authors estimate a panel regression in which they constrain changes in relative prices over time to be explained only by distance from downtown. In our exercise, by re-estimating the equation 32 times (one estimation per quarter), we allow estimated coefficients on all neighbourhood characteristics, including distance, to change over time. Nevertheless, our findings remain qualitatively similar when we apply the methodology from Gupta et al. to the Canadian data.[]
7. 7. Note that the CMA variable may capture differences in supply elasticities or other supply factors between the various downtown cores in Canada. However, this variable would not account for differences in supply within each sub-area of a given CMA.[]
8. 8. Note that 80 kilometres may not imply the same level of urbanization across the various CMAs. The key results are similar whether we set the distance threshold to 60 or 100 kilometres.[]
9. 9. We consider other control variables in the exercise, but they were not statistically significant. These include (all fixed at their 2016 Census values) the share of the population living alone, the share of population employed whose commute time is more than 45 minutes, the population density (population per square kilometre), the employment rate, the share of the population with low income, the share of renters, the average number of rooms per dwelling, whether the FSA has a rural status (second character of the FSA is a zero), and the share of population employed in industries with jobs that are conducive to working from home (based on the methodology of Dingel and Neiman 2020).[]
10. 10. As a robustness check, we estimate the equation with the average qualifying income reported by financial institutions when they issued a new mortgage (which varies over time). Results were similar.[]
11. 11. Estimating the equation without the control variables specific to each FSA leads to an estimated bid rent gradient that is between 1.5 to 2 times more negative. This implies that about 40% to 50% of the proximity premium can be accounted for by neighbourhood characteristics.[]
12. 12. For a discussion of supply elasticities in Canadian cities, see Paixão (2021).[]

## References

1. Alonso, W. 1964. Location and Land Use: Toward a General Theory of Land Rent. Cambridge, MA: Harvard University Press.
2. Dingel, J. I. and B. Neiman. 2020. “How Many Jobs Can Be Done at Home?” National Bureau of Economic Research Working Paper No. 26948.
3. Gupta A., V. Mittal, J. Peeters and S. van Nieuwerburgh. 2021. “Flattening the Curve: Pandemic-Induced Revaluation of Urban Real Estate.” Journal of Financial Economics (November 2).
4. Paixão, N. 2021. “Canadian Housing Supply Elasticities.” Bank of Canada Staff Analytical Note No. 2021-21.

## 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, R23, R3, R32

DOI: https://doi.org/10.34989/san-2022-7