Appendix A

The next step was to use these analysis results to simulate how the headship rates "should" have changed during 1996 to 2001, given the changes that occurred in the economic variables. In other words, for each of the 25 CMAs, for each of the age groups, the predicted changes in headship rates were compared to the changes that actually occurred. Figure 32 shows that:

• For the three younger age groups, the economic analysis predicts relatively large increases in the headship rates, because homeowners' costs fell during the period (in inflation-adjusted dollars) and the employment-to-population ratios increased in most of the CMAs (due to economic expansion in most areas of the country). In actuality, however, headship rates for the younger age groups changed very little or fell. Therefore, the predictions generated for the younger age groups were far off the mark.
• For the three older age groups, the predictions were more accurate, as increases were predicted and increases actually occurred.

Figure 32: Predicted vs. Actual Changes in Headship Rates, for 25 CMAs, by Age Group, 1996 to 2001

In conclusion, this analysis did not generate results that could be used to confidently predict future headship rates. However, the analysis does support a theory that economic variables do affect household formation rates, and therefore that changes in future economic conditions are likely to influence headship rates. The directions of the influences are:

• A rising cost of homeownership is likely to reduce household formation and a falling cost to increase household formation.
• A rising employment-to-population ratio is likely to increase household formation and a falling ratio to reduce household formation.

If conclusions can be drawn about the likely directions of these economic variables in future, then conclusions could also be drawn about the directions of headship rates.

Analysis of Household Sizes

In addition, data on the distribution of the adult population by age was included, since different age groups may have different household sizes.

Various combinations of the data were tested. The model that best explains variations in the number of adults per household uses average owners' costs and the employment-to-population ratio, as well as data on age distributions: the two economic variables that were most influential in this analysis also produced the most reliable results in the headship rate analysis.

Figures 34 and 35 show the relationships between household sizes and the two key economic variables. In Figure 34, there is a clear relationship between the owners' costs and the number of adults per household, and the relationship is quite strong, as indicated by the clustering of most of the data points close to the trend line and the relatively high R-Square figure of 0.545. Moreover, the relationship has the expected direction -- higher housing costs are associated with larger household size.

Figure 34: Number of Adults per Household vs. Owners' Monthly Costs, for 25 CMAs

Source: Will Dunning Inc., using data from Statistics Canada 1996 and 2001 Census of Canada

Table 6.5-2 shows the analysis of that shortfall.

Three and a half years have passed since the 2001 census, and if the "expected" adjustment is occurring, it may be possible to find evidence of the adjustment, by estimating the extent to which household sizes in the GTA have changed in the interval.

Much has changed since the 2001 census. As of December 2004:

• Housing completions have added an estimated 162,576 new dwellings.
• Vacancies have increased. Using CMHC data on vacancies in the conventional apartment stock and adding an estimate of vacancy change in the rental stock not covered by CMHC's survey, the number of vacant units is estimated to have increased by 18,229 units.
• Subtracting the change in vacancies from the number of completed units, the number of occupied dwelling units (households) is estimated to have increased by 144,347, or about 40,300 units per year.
• The adult population of the GTA has expanded, according to Statistics Canada's Labour Force, by 368,500. This would result in a total adult population of 4,450,910 as of December 2004.
• Thus, as of December 2004, it is estimated that the average number of adults per household in the GTA is 2.312. This is an increase of 0.020 adults per household from the 2001 census figure.

Thus, while we were looking for evidence that household size in the GTA is falling, this analysis suggests that it has actually continued to increase.

In terms of the economic variables:

• The monthly cost of homeownership has increased. According to the Consumer Price Index, the cost of owned accommodation in the Toronto CMA increased by 10.5 percent during May 2001 to December 2004. The overall inflation rate was 6.7 percent. Therefore, the cost of homeownership increased by 3.5 percent in inflation-adjusted terms. This should tend to increase household sizes.
• The employment-to-population ratio has fallen. In both the Toronto CMA and the GTA, the seasonally adjusted employment rate fell by about 0.9 from May 2001 to December 2004. This should also increase household size.

Combining these economic effects with anticipated shifts in the age structure of the population, the average number of adults per household in the GTA should have increased by a small amount -- 0.007 adults per household -- and the average number of adults per household should be about 2.299. The actual increase was larger than predicted (0.020 people, almost triple the predicted amount) and the average household size is larger than predicted, at an estimate of 2.312 versus the predicted 2.299.

Conclusions