Notes

^1.

The years covered by the Famex are 1969, 1974 (urban areas only), 1978, 1982, 1986, 1990 (urban areas only), and 1996.

^2.

While there is continuity in these surveys, analysts must always be cognizant that a particular year may offer less accuracy than others—either because the sample drawn for the survey of that year was less representative than the sample drawn for surveys of other years or because there were other problems that year. Comparisons across years such as presented here provide hindsight on potential incongruities.

^3.

Since the 2001 SHS includes only house and condominium values for households that bought and/or sold their homes in the reference year, self-reported house and condominium values are incorporated using the 2001 Census of Population. To do so, the median house values are obtained by census subdivision, further broken down by age class. The timing of the questionnaire for the 2001 SHS and the 2001 Census of Population are not the same; consequently, the self-reported housing value may not completely match the data in the SHS as a result of this lag.

^4.

The 1978 Famex does not contain this information.

^5.

See Myles (2000) for a discussion of the relative contributions of private pensions, investment income, OAS/GIS, and CPP/QPP to the growth in incomes of the population above age 65 from 1980 to 1995.

^6.

For instance, Myles (2000) finds that private pension plans accounted for an increasing share of seniors' incomes from 1980 to 1995.

^7.

The 1969 and 1974 Famex do not include the number of bathrooms in the home.

^8.

The model does not include a variable that controls for size of the urban area of the dwelling because the definitions of area size were not consistent over time and no such variable was present in the 1990 Famex. To test the sensitivity of the estimates to the inclusion of size of urban area, it was included in the 1982, 1986, and 1996 models. In comparison to the original estimates, the outcomes were very similar; therefore, in order to be consistent over time, the model does not include a variable that controls for size of the area.

^9.

The imputed rents reported herein are corrected in order to take into account the log-linear nature of the model. This correction does not have a qualitative effect on the results. It is not made in Brown, Hou and Lafrance (2010), which relied primarily on the user cost-based measure of housing services.

^10.

Detailed information is included only post-1992. However, the costs as a share of contract rents do not vary much over time. Consequently, for the years prior to 1992, a constant share is applied.

^11.

Since rental values are not available in the 1978 Famex, r_t - a_t is obtained in 1978 by taking the average r_t - a_t from 1974 to 1982.

^12.

For condominiums, a portion of repairs and maintenance is included in condominium fees. Therefore, 10% of condominium fees are added to expenditures on repairs and maintenance, except for the years 1969 to 1978, for which condominiums cannot be identified.

^13.

For condominiums, a portion of expenditures on improvements and alterations are included in condominium fees. Therefore, 30% of condominium fees are added for improvements and alternations, except for the years 1969 to 1978, for which condominiums cannot be identified.

^14.

Since the 1990 Famex only surveyed households living in urban areas, the value of housing services may be biased upward in that year since house prices, on average, tend to be higher in urban locations.

^15.

This conclusion is expressed with caution because definitions and changes in survey practices may have contributed to these changes.

^16.

See Courchane and Giles (2002) for a more in-depth discussion of trends in the Canadian and U.S. mortgage markets.

^17.

Here estimates of returns to equity take into account only the opportunity cost of capital plus repairs and maintenance as well as depreciation, less the expected appreciation in house prices.