See Table 7.

This probably overstates the true opportunity costs slightly since the actual mortgage rate in a transaction is below this posted rate.

Only the housing-service component of expenses is of interest here—not other services, such as utilities, which are conceptually different from the space rent.

A measure of depreciation could have been derived from that used by the CPI (1.5%); however, that estimate was derived from somewhat dated U.S. studies and would not have been consistent with the other expenses derived from the SHS.

These expenditures are made in order to increase the value and useful life of the property. They include the following: structural additions or extensions; remodeling rooms; replacements and new installations of fences, driveways, patios, swimming pools, major landscaping, roofs, eaves troughing, exterior walls, windows, doors, hard-surface flooring, wall-to-wall carpeting, plumbing fixtures and equipment, heating and/or air-conditioning equipment, electrical fixtures and equipment, and built-in appliances; and other improvements and alterations.

For condominiums, a large portion of expenditures on improvements and alterations are incorporated into condo fees. Therefore, 30% of condo fees are added to expenditures on improvements and alterations.

For instance, the replacement of siding on a townhouse will be less than that for a detached home because a townhouse will have at most three sides.

Alternately, we could have used the rate of 1.5% used in the CPI. But the latter takes into account both structural and non-structural depreciation, and is based on studies that are dated. We felt it better to directly estimate the non-structural component and then to add an estimate for the latter. See Buergel-Goodwin, Ferrara, and Bradley (2005) for evidence that the structural component depreciates at very low rates.

Harding, Rosenthal, and Sirmans (2007) argue that depreciation gross of maintenance in the US between 1981 and 2001 averaged around 2.5%. The estimate used here for depreciation plus maintenance is 2.43%.

SHS Data Dictionary, 2006: Data Model Entity (Internal Household).

For condo, 10% of condo fees are added to expenditures on repairs and maintenance.

Using housing price data dating back to the late nineteenth century, Shiller (2007) finds, during the post-2000 period, significantly higher house prices than experienced over a century of data, for the United States, Norway and the Netherlands.

This method is meant only to smooth out the estimates of user costs—not to test whether a housing bubble existed at any point in time.

This occurred only when the terms of trade were included in the model.

For instance, the contribution to utility of an additional bedroom in a home will depend on the number of bathrooms in that home.

The SHS also provides information on major and minor repairs; however, by construction of the rent variable, dwellings in need of major repairs are excluded. Inclusion of the period of construction of the dwelling in the SHS model had no significant effect on results; consequently, it was dropped from the final specification reported in Table 3.

Neighbourhoods are defined as dissemination areas.

We experimented with fixed effects for dissemination areas in order to better account for location within urban areas and other unobserved neighbourhood characteristics (e.g., access to recreational opportunities or good schools). However, this had only a modest effect on our results, and led to complications when predicting rents for owner-occupied homes in disseminations areas with no or very few rental housing units.

Higher rental rates for apartment buildings over five stories in height may reflect the generally more central location of these types of rental accommodation.

While the negative effect of owned dwellings might be considered an unexpected result, it is important to keep in mind that a higher share of owner-occupied housing may be correlated with a location on the outskirts of metropolitan areas, where rental rates would tend to be lower in order to compensate for higher (pecuniary and non-pecuniary) commuting costs.

Alternate sensitivity tests can be performed with these ranges. For example, if it is assumed that capital costs are 1% lower than employed here (possibly because some mortgages were written at a discount or used variable, rather than fixed, rates), this corresponds to the results produced under the 6% column. Unpublished research within Statistics Canada suggests a 1% discount on the capital costs is not unreasonable.

If quintile ranges from the SHS are utilized, which are somewhat higher than those from the Census, the number of observations in the top quintile falls to 69, and the average rent paid for these homes rises to $17,900.

This may reflect lower house prices reported in the Census of Population; these are due in part to the earlier reporting data. The 2006 SHS survey was conducted between February and April 2007 and asks respondents the value of their home at that time, while the 2006 Census of Population was conducted in May 2006 and asked respondents to estimate the value of their home at the time of the Census. Consequently, there is approximately one year between the two surveys. It is also possible that the Census of Population provides a better sample coverage of low-value homes.

It should be noted that this is also the result when one assumes a longer-run price average for expectations and a discount from the fixed-rate mortgage rate.

Returns to equity are calculated by multiplying the equity share of the home by the value of housing services accounted for by equity. This is the cost-of-capital portion of the user-cost measure of housing services, and so takes into account the opportunity cost of capital plus depreciation less expected appreciation in house prices.

The same pattern holds when homeowners and renters are combined.