3.0 Who is more likely to spend 30% or more of income on shelter costs?

View the most recent version.

This section explores this question using two regression models. The first compares the characteristics of those who ever spent 30% or more of their household income on shelter costs for at least one year during the study period to those who never did. The second model compares those who persistently (for all three years) spent above the benchmark to those who had at least one year where they were below it (the never plus occasionally group as per table 2). Appendix D provides more information on the distribution of the sample for each of these models and Appendix E has information about the variables used in the models.

3.1 Average Canadian has one in five probability of living in a household spending 30% or more for shelter

The first model predicts that the average¹ Canadian has a probability of about one in five (21%) of ever living in a household spending 30% or more of income on shelter over the three-year study period. However, the probability (based on the second model) of this average Canadian living in a household that persistently exceeds the affordability benchmark is much lower at 3.9%.

Note that these probabilities differ from the percentages presented in table 6 in the previous section. Table 6 presents the percentages of the population with various characteristics that exceeded the affordability benchmark during the study period. On the other hand, table 7 presents the probabilities of ever or persistently exceeding the affordability benchmark for populations with various characteristics if all other characteristics (as included in the model) are held constant. The model helps us to isolate the effect of a single characteristic at a time. See Appendices D, E and F for more information about the models.

Individuals and families living through changes affecting their incomes or shelter costs see corresponding changes in their STIRs and hence, their probabilities of ever or always spending 30% or more of their incomes on housing. Movers, those who change tenure, and those whose family situation changes (perhaps through divorce, marriage or other family changes) are particular examples of those whose circumstances have changed and they are featured in the discussion below.

Age is not a strong factor in determining the probability of ever spending 30% or more on shelter costs. None of the age groups in the first model had probabilities significantly different from the reference category (age 30-49) of ever having a STIR of 30% or more. For the second model, only the 20-29 age group had a significantly lower probability from those aged 30 to 49 of persistently exceeding the affordability benchmark. However, even though this difference was significant, it was not substantially lower. Perhaps there are a variety of reasons why the 20 to 29 year age group was significantly different from the reference category. This is a group in transition. Some still live at home with their parents and therefore their shelter costs and income reflect their family's situation rather than their own. Those who have moved out may be saving to buy a house and live in inexpensive accommodation to do so. If they have not yet started a family, they will not need the larger, more expensive accommodation required by families. Note that, while many in this age group share accommodation with roommates, these households are excluded from this study and this should be kept in mind when interpreting the results presented here.

3.2 Subsidized renters have a significantly lower probability than market renters of persistently spending 30% or more for shelter

Past Census results have shown that on average, renters have higher STIRs than owners and are more likely to spend 30% or more of their incomes on shelter. Since this study looks at a three-year period, the typical owner-renter groupings have been expanded. The tenure variable used in this report groups owners by whether they have mortgages or not, and renters by whether they pay market or subsidized rent. Persons in each of these four groups did not change their situations during the full three-year period. Separate categories are provided for owners who added or discharged a mortgage, for renters who moved from paying market rent to subsidized rent or vice versa, and for those who changed from renting to owning or vice versa. Table 7 shows the share of the study population living in each of these tenure situations.

The reference category in the model is renters who paid market rent all three years. Considering previous experience with Census information, we would expect owners generally to be in more affordable situations, and to be less likely than market renters to have STIRS of 30% or more. Similarly, we would expect mortgage-free owners to be in more affordable situations than those with mortgages. And we would anticipate that subsidized renters would be in more affordable situations than if they were paying market rent.

The model results support our expectations and allow for a more in-depth understanding of the role that tenure plays in housing affordability. In table 8, the tenure categories are arranged in descending order based on their median STIR.

Renters have the highest median STIRs. They are also the most likely ever to spend above the affordability benchmark and, with the exception of subsidized renters, persistently to spend above this benchmark. The results of the first model show that, for market renters, the probability of ever spending 30% of household income or more on shelter is one in three. The probability drops to one in eight for persistently spending above this benchmark and is even lower for subsidized renters at one in eighteen. This is in spite of the fact that the median income for subsidized renters is only half that of market renters. Thus, the results of the second model provide additional evidence that rent subsidies are having an effect.

While it may seem counterintuitive that subsidized renters have higher STIRs than market renters, they would be much higher without rent subsidies. If subsidized renters had paid the median market rent of $8,301 rather than their subsidized rent, their median STIR would have been 42% instead of 26%. The median shelter costs of renters subsidized for all three years are 40% below those of market renters. This helps to make their shelter costs much more affordable when compared to their very low median incomes.

Also consistent with expectations, owners without mortgages have the lowest STIRS. When examining the probability of ever spending 30% or more of household income on shelter, at only 5.2%, owners without mortgages are in a class of their own, far below the 26% of the next lowest tenure group of Canadians, owners changing their mortgage status.

While the 36% of Canadians who are owners with mortgages have about the same probability as market renters as a group of ever exceeding the affordability benchmark, they do so under totally different circumstances. Table 8 shows that owners with mortgages had the highest median incomes at $79,306, and also the highest median shelter costs, at $15,282. Having the highest incomes, owners with mortgages are better able to afford to pay a higher percentage of income on shelter. The high payments of owners include mortgage interest as well as mortgage principal.² In contrast, market renters had median incomes which were only half those of owners with mortgages, but median shelter costs that were more than half.

While the nearly 10% of households that changed tenure during the three year study period had a relatively high probability of ever spending 30% or more of income on shelter, they had a lower probability of doing so on a persistent basis. The reasons for this are likely complex. It may be that their tenure change is associated with short term high STIRs but that, in the longer term (in this study, over a period of three years), their situation improves. The change in tenure may be associated with such varied circumstances as a move that temporarily increases shelter costs faster than income or a move to adjust to family breakup and a drop in income.

3.3 Toronto and Vancouver residents stand out

"Location, location, location", so often heard in real estate, can also be used about shelter costs. Housing costs are highest in Canada's largest metropolitan areas. Are these higher costs reflected in higher probabilities of exceeding the housing affordability benchmark in these cities? The model suggests that Canadians living in Vancouver and Toronto, two of the largest and the two most expensive cities in Canada, have significantly higher probabilities of ever and persistently exceeding the affordability benchmark compared to people living in Ottawa-Gatineau, the reference category.

Table 9 presents median household STIRs, shelter costs and incomes for the metropolitan areas in this study. It shows that Torontonians shoulder the highest median shelter cost of all metropolitan areas, but do so on one of the highest median incomes, which mitigates their STIRs, leaving Torontonians with the second highest STIR. At 21%, it is Vancouverites that have the highest median STIR.

Canadians living in Montreal, Calgary, Edmonton, other CMAs and rural areas have about the same probability of spending above the housing affordability benchmark as Ottawa-Gatineau residents. Residents of Victoria, accounting for the smallest population share of all the centres in the study, have a relatively high probability (28%) of ever spending 30% or more of household income on shelter. However, in terms of persistently exceeding the affordability benchmark, they are not significantly different from Ottawa- Gatineau.

Finally, as noted earlier, people living in households whose circumstances change tend to have higher STIRs. Results show that Canadians who move between metropolitan areas have a significantly higher probability (28.0%) of ever exceeding the affordability benchmark than those living throughout the study period in Ottawa-Gatineau (the reference category). But in terms of always exceeding the benchmark, results show that movers are not significantly different. Possibly it just takes time to find a good job and affordable shelter in a new city.

3.4 Family-related transitions an important factor in housing affordability

Family living arrangements are not static. Various events can result in changes in family composition, for example, marriage, divorce, separation, death, and the departure or return of grown-up children. Between 2002 and 2004, 13% of the population changed from one family type category to another. In order to study and compare families that changed to those that did not, a separate category was created.³

Table 10 shows median shelter costs and incomes for each family type, ordered by the value of the shelter cost-to-income-ratio. Female lone-parent families have the highest STIRs at 27%, followed by women and men living alone. All three of these groups have median incomes that are less than half of that of couple families. Those living alone have median shelter costs that are less than two-thirds of those of couples, but female lone-parents pay almost 80% of what couples pay for shelter – which is why their STIRs are the highest.

Couple families, the most common family type, account for 66% of all people in Canada. For couples, the probability of living in a household ever spending more than the affordability benchmark is 16% and the probability of doing so persistently is just 3%, both comfortably below the national average. Couple families benefit from having the highest median income of all family types which offsets their high shelter costs, giving them almost the lowest median STIR.

In contrast, those living in the remaining family types are significantly more likely ever to spend 30% or more of income on shelter – especially female lone-parent families and women living alone. Families in these two categories also have the highest probabilities of persistently spending above the benchmark. These smaller families are not able to benefit from the possibility of having more than one income (whether from government transfers or a salary). Perhaps even more important, the average earnings of employed women are still substantially lower than those of men, even when they are employed on a full-time basis. In 2003, women working on a full-time, full-year basis had average earnings of $36,500, or 71% what their male counterparts made.⁴

Those whose family type changed deserve special mention. Like those who moved or changed tenure, their probability of ever exceeding the benchmark is elevated. It is much higher than for couples, though not as high as for women living alone and female lone-parents Their probability of always exceeding the benchmark is also significantly higher than for couples, though not by much. Perhaps people in families who add or lose members are able to make adjustments that reduce their STIRs after a year or two, whereas women living alone or bringing up children by themselves do not have such flexibility.

3.5 Recent immigrants and visible minorities have high probabilities of ever spending more than the housing affordability benchmark

More than 70% of immigrants who arrived in Canada since 1982 belong to a visible minority group.⁵ For this reason, the findings for recent immigrants and visible minorities are discussed together.⁶

The high proportion of recent immigrants who are also visible minorities is not the only pertinent similarity between these two groups. Both also tend to live in Canada's largest urban centres where shelter costs are highest. For example, in 2001, 86% of immigrant households compared to 58% of non-immigrant households lived in Canada's census metropolitan areas (CMAs).⁷ Both groups are more likely than Canadians in general to live in Toronto and Vancouver. In 2001 the share of visible minority Canadians living in households in these two CMAs was four times larger than the share of Canadians who were not part of a visible minority. Just over 40% of all visible minorities compared to 11% of Canadians who were not visible minorities lived in Toronto while in Vancouver the comparable percentages were 18% and 5%.⁵

Another similarity between visible minorities and recent immigrants is the size of their families. The average size of visible minority families in 2002 was 3.8 compared to 2.9 for families that were not visible minorities. Similarly, the average size of recent immigrant families was 3.7 for those in Canada less than 10 years and 3.9 for those in Canada for 10 to 19 years. For the Canadian-born, the average family size was lower at 3.0 (tables 11 and 12). Larger families require larger accommodations, pushing up shelter costs. However, larger families can also generate more income, through the efforts of additional earners or from transfer payments. In fact, tables 18 and 19 show that, while median household incomes are similar for those who are visible minorities and those who are not, and for immigrants (except for the very recent ones) and Canadian-born, shelter costs are much higher for visible minorities and recent immigrants.

Given their tendency to live in the largest, most expensive cities and their larger family sizes, it is not surprising that both recent immigrants and visible minorities have significantly higher probabilities of living in households spending 30% or more of income on shelter at least once during the three-year study period. For immigrants (including visible minorities and those who are not), this higher probability declines with the length of time they have lived in Canada, with those who have spent 40 years or more in Canada having probabilities not significantly different from the Canadian-born. Immigrants who had lived in Canada for less than 10 years had the highest probability of ever exceeding the affordability benchmark at 39%. This probability drops to 23% for those in Canada for 40 or more years.

Results are similar for immigrants persistently exceeding the affordability benchmark. Recent immigrants are significantly more likely than the Canadian-born to exceed the benchmark and this probability drops off with the number of years in Canada until there is no significant difference from the Canadian-born. For visible minorities however, there is no significant difference between them and those who are not visible minorities in the probability of persistently exceeding the affordability benchmark.

Thus, locating in Canada's largest cities and living in larger families can increase shelter costs for visible minorities and recent immigrants. It can also contribute to higher incomes since cities offer more job opportunities and larger families can generate more income. For visible minorities, the probability of ever exceeding the affordability benchmark is significantly higher than for non-visible minorities, although the probability of persistently exceeding it is not. For recent immigrants, both probabilities are elevated compared to the Canadian-born, especially for the most recent immigrants who had been in Canada for less than 10 years at the beginning of the study. These most recent arrivals are also the group whose income is quite a bit lower than other immigrant groups. If they go on to follow the pattern of immigrants who came before them, they will gradually increase their incomes and reduce their shelter costs and family sizes. This in turn would reduce their STIRs and their probabilities of exceeding the affordability benchmark. Future studies should examine this group more closely.

3.6 Aboriginal households are more likely to spend more than the affordability benchmark, but not persistently

Unlike immigrants, Aboriginals living off reserve⁸ do not congregate in Toronto and Vancouver. Only 11% of Aboriginal Canadians lived in these two CMAs compared to 22% of non-Aboriginals. This difference in location likely accounts for the lower median shelter costs they report, $8,286 compared to $9,088 for non-Aboriginal Canadians. But their lower shelter costs are associated with even lower incomes resulting in STIRs that are higher than for non-Aboriginals (table 13).

Results from the two models show that Aboriginals living off reserve are significantly more likely than non-Aboriginals ever to exceed the affordability benchmark, but no more likely to do so persistently. Aboriginals have a higher rate of moving than non- Aboriginals. The percentage of Aboriginals that moved during the three year study period is 17.4%, compared to 12.4% for non-Aboriginals, and, as already seen; households that move are more likely to exceed the affordability benchmark.

Results discussed above showed that Canadians moving between metropolitan areas had a significantly higher probability of ever exceeding the affordability benchmark but, in terms of always exceeding the benchmark, they were not significantly different. Perhaps it is the higher mobility of Aboriginal Canadians that causes them to have a similar pattern to movers. As noted above, it may take time to find a good job and affordable shelter in a new city.

On the other hand, other characteristics of Aboriginal housing include higher rates of crowding or unsuitable housing as measured by the National Occupancy Standard⁹ and higher rates of living in a dwelling unit in need of major repairs. See table 14. It may be that Aboriginals are living in inadequate or unsuitable accommodation to lower their rents. This is an area that merits further study.

3.7 Disabled more likely to exceed affordability benchmark

Those who self-identified as disabled at least once during the three year study period have significantly higher probabilities than the non-disabled of ever or always exceeding the affordability benchmark. The disabled are also more likely to live in families where the major source of income comes from government transfers (including old age security) and not wages and salaries (table 15).

Table 16 Median shelter-cost-to-income-ratio (STIR), shelter cost and income for the disabled and the non-disabled, 2002

3.8 Higher education linked to higher earning power and lower STIRs

As would be expected, compared to those who received some kind of post-secondary certification other than a BA, those with less education have significantly higher probabilities of ever or persistently exceeding the affordability benchmark. Similarly, those with more education (BA, or post graduate degrees) have significantly lower probabilities of doing so. (table 17).

____________

1. By setting all model variables to their mean value, we can mimic an "average" Canadian in the sample.
2. The principal portion of a mortgage payment helps to build equity and therefore household wealth. The breakdown of mortgage payments into principal and interest is often not known by respondents and is not asked on SLID. Owners with mortgages who spend 30% or more of their income on shelter (i.e., they do not meet the affordability benchmark) are, unlike renters, contributing to their wealth.
3. Family type categories used in the models are: couple families, female lone-parents, women living alone, men living alone, other family type and changed family type. Categories are assigned to individuals based on all members of the family, even though children aged less than 16 are not included in the models. Also, households with more than one economic family are not part of this study. Couple families include those with and without children (under 18). This category includes married, common-law and same-sex relationships. Female lone-parent families include at least one child and the mother must be younger than 65. The category "other family type" includes male lone-parent families and couples or lone-parent families with other relatives living with them. Those who changed family type during the three-year period could have married, separated or divorced, had a death in the family, had children turn 18, or had relatives (including children 18 or over) leave home or take up residence with them.
4. Women in Canada: A Gender-based Statistical Report, 2005 Catalogue No. 89-503-XPE.
5. Canada Mortgage and Housing Corporation. forthcoming. 2001 Census Housing Series: Canada's Visible Minorities. Research highlights Socio-Economic Series, forthcoming. Ottawa: Canada Mortgage and Housing Corporation.
6. Immigrants are those born outside Canada and who have been given the right to live in Canada permanently by immigration authorities. Visible minority status is defined based on three questions: mother tongue, ethnic or cultural group of ancestry, and country of birth. Recent immigrants are defined based on the "years since immigration" variable. For this report, those who immigrated in the 20 years before 2002 are recent immigrants. Those who immigrated in the 9 years before 2002 are the most recent immigrants.
7. Canada Mortgage and Housing Corporation.(2004). 2001 Census Housing Series: Immigrant Households, table 2, p. 2. Research highlights Socio-Economic Series 04-042 Ottawa: Canada Mortgage and Housing Corporation.
8. Those in the Aboriginal category indicated at least one of the following: that they were a Treaty Indian or a Registered Indian as defined by the Indian Act of Canada; or that their ancestors were Cree, Micmac, Métis or Inuit. This method of defining Aboriginal is different from the Census. In the Census an identity approach is taken and those in the Aboriginal category answered yes to at least one of the following: that they were an Aboriginal person; that they were a member of an Indian Band or First Nation; or that they were a Treaty Indian or a Registered Indian as defined by the Indian Act of Canada. The SLID definition gives a higher estimate for off reserve Aboriginals: 629,000 (aged 16 and over) in reference year 2001 compared to 471,000 (aged 15 and over) on the 2001 Census. The reason for SLID's higher estimate is that it includes those with Aboriginal ancestry.
9. Overcrowded dwellings do not have enough bedrooms for the size and make-up of resident households, according to National Occupancy Standard (NOS) requirements. Enough bedrooms based on NOS requirements means one bedroom for each cohabiting adult couple; unattached household member 18 years of age and over; same-sex pair of children under age 18; and additional boy or girl in the family, unless there are two opposite sex siblings under 5 years of age, in which case they are expected to share a bedroom. A household of one individual can occupy a bachelor unit (i.e. a unit with no bedroom). The NOS, developed by a Federal/Provincial/Territorial Working Group to be the suitability standard for use in the core housing need model, embodies the common elements of standards in place by different provincial and territorial housing agencies across Canada. For more information on the role of the NOS in defining acceptable housing and in the measurement of core housing need, see Core Housing Need in Canada, 1991, Canada Mortgage and Housing Corporation. Pgs. 3-7.