Income Research Paper Series
Low Income in Canada - A Multi-line and Multi-index Perspective
Low income across groups of people
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
In Chapter 2, we drew a picture of low-income Canadians living in the ten provinces. We found that low income improved considerably over the 2000 to 2009 period when it was measured by fixed standards. Also, the depth and severity of low income declined from 2000 to 2007, even under a variable standard, but there was evidence to suggest that an upward trend in low income may have started to emerge in 2008. Do these findings accurately describe the developments of well-being for different groups of people? This chapter attempts to answer this question.
People can be classified in several groups. When classified by age, low income among children and seniors may be examined. When classified by family status, low income among persons living in lone parent families (lone parents) and unattached people may be investigated. When classified by race and immigration status, low income among new immigrants and off-reserve Aboriginal people may be examined. This chapter provides results of investigation into low income based on classifications by group.
There is significant interest in the well-being of children in Canada, so we begin by looking at low income among children.1Figure 3.1 contains the estimated low-income rate and gap ratio for Canadian children from 1976 to 2009. Low income among children can be characterized by two periods of upward movement and two periods of downward movement. From the late 1970s to the early 1980s, the low-income rate increased by about 3 percentage points under both the low-income cut-off (LICO) and the low-income measure (LIM). This was followed by a period in which the low-income rate declined by 3 percentage points under LIM and 4 percentage points under LICO. The second period of upward movement started in the late 1980s and lasted until the mid-1990s. This was followed by a prolonged period of improvement in low income that lasted about 15 years, to 2007. The decline in low income among children was particularly strong according to the two fixed standards, LICO and the market basket measure (MBM).
Low income among Canadian children has several interesting characteristics. First, low income among children, at times, was rather high. For example, in 1997, the low-income rates for children were as high as 17% and 18% under LIM and LICO, respectively, much higher than the 13% and 15% experienced by working age adults. However, overall the low-income rate among children was lower than the rate for people in several other vulnerable groups.
Second, low-income measures among children under different lines were quite close to each other until the late 1990s. Moving into the 2000's, different lines gave somewhat different results. In particular, low income under the LIM was stable during the 2000 to 2009 period, while low income under LICO and MBM continued to drop during this period. Under LICO, the low-income rate dropped by five percentage points, from 14% in 2000 to 9% in 2009, while under MBM, it dropped by four percentage points, from about 14% in 2000 to 10% in 2008. But under LIM, the low-income rate for children was basically unchanged over the same period.
Figure 3.1 Low-income rates (top panel) and gap ratios (bottom panel) for children
Third, the short-term variations in low-income rates under different lines from 2007 to 2009 drew our attention. From 2007 to 2008, the low-income rate under LICO dropped by half a percentage point. However this is likely not indicative of an increase in the economic well-being of Canadian children during that time. On the one hand, the decline in the low-income rate was not statistically significant.2 On the other hand, the estimated gap ratio (bottom panel, Figure 3.1) and severity (not shown here) under LICO actually increased slightly. In other words, the "decline" in the low-income rate under LICO was not supported by movements in the gap ratio and severity indexes under the same line.
In contrast, the number of children in low income under MBM and LIM went up in 2008. This upward movement continued in 2009 under MBM while falling back slightly under the LIM. As can be seen in Figure 3.1, the low-income rate for children under MBM increased from 9.4% in 2007 to 10.3% in 2008 and to 11.6% in 2009, with low income rate in 2009 being higher than that in 2007 with marginal significance.
There has been a long term decrease in low income among seniors (65 years old or older) since the late 1970s. Figure 3.2 shows that dramatic declines in low-income rates for seniors occurred from the late 1970s to the mid 1990s under both LICO and the LIM. We also detected similar trends in the low-income gap ratio and severity indexes for this period under LICO and LIM. However, the low income trends under different lines started to diverge in the mid 1990s. Under LICO and later under MBM, low income continued to follow a generally declining trend. Under LIM, however, low income started to follow an upward trend.
Nevertheless, the upward trend in the low-income rate under LIM is probably not indicative of a serious worsening of the low income situation for seniors. On one hand, the low-income rate under LICO continued to decline in recent years. For example, under LICO, the low-income rate for seniors declined from about 10% in 1995 to about 5% in 2007. On the other hand, our calculations (not shown here) suggest that other low-income indexes were relatively low, no matter which low-income line was employed. The gap ratios were always under 2%, while the severity indexes were generally below half a percentage point, far below those for the overall population (Figures 2.7 and 2.8).
However, the development in seniors' low income between 2007 and 2009 is worth noting. Compared with 2007, seniors' low-income rate increased under both LIM and MBM in 2009. The increase under LIM was in the neighbourhood of one percentage point, and the change was marginally significant. Under MBM, however, the low income rate doubled between 2007 and 2009, from 2.1% to 4.2% (a significant increase). These changes suggest that the low-income rate among seniors is more sensitive to economic fluctuations than among other groups because, as indicated by Osberg (2001), there are spikes in seniors' income distribution around the low-income threshold such that a small change in their income or the threshold may result in a large change in the incidence. This is also probably the reason why the changes in low-income gap ratio or severity were small under all thresholds.
Figure 3.2. Low-income rates among seniors 1976 to 2009
Since 1995 the increase in the low-income rate for seniors under LIM indicated that their income did not rise as quickly as the income of non-seniors. As can be seen from the top panel of Figure 3.3, the evolution of income (after-tax and transfers) of seniors and non-seniors from 1976 to 2009 can be divided into three stages. From 1976 to 1989, the income of seniors and non-seniors followed a similar path. From 1990 to 1997, incomes of both groups grew slowly, while the income of seniors grew slightly faster than that of non-seniors. But from 1998, the paths of income for seniors and non-seniors started to diverge, with income of non-seniors growing at an increasingly higher rate than that of seniors.
One possible factor behind the slower growth of seniors' income in the third stage was the slowed growth of government transfers to seniors. The lower panel of Figure 3.3 shows the growth of total government transfers to senior Canadians during the 1976-to-2009 period. Starting from the early 1990s, the median government transfers to seniors increased at a slower rate relative to the period before the early 1990s. Indeed, from 1976 to 1994, the annualized growth rate of median government transfers to seniors was 8.7%, while from 1995 to 2009, the annualized growth rate was 2.0%.
Low income among lone parents in Canada has been recently examined by Richards (2010). He characterizes the remarkable decline in the low-income rate for lone parents as a success story. This report adds two insights to his work.3
First, we examined higher order low-income statistics, in addition to the low-income rate, for lone parents. As can be seen from the top panel of Figure 3.4, the low-income rates among lone parents fluctuated from 40% to 50% for about 20 years, that is, from 1976 to 1996, according to both the LICO and the LIM lines. Richards (2010) observed that a long-term declining trend in the low-income rate started in 1997 and it declined further to a historically low level in 2007. However, the gap ratio and severity indexes evolved somewhat differently from the low-income rate. The bottom panel of Figure 3.4 suggests that the depth of low income started a long-term declining trend long before 1997, when the low income rate started to decline. We also calculated the severity indexes under different low-income lines for the period from 1976 to 2009 and found the trend to be similar to that of the gap ratio.
Figure 3.4 Low-income rates (top) and gap ratios (bottom) for lone parents
Second, despite the slight increase in low income for the general population from 2007 to 2009, low income for lone parents did not increase in that period. Rather, it continued along a declining trend that started earlier (1997 in the case of the low-income rate and 1986 in the case of the low-income gap ratio and severity) under LICO and LIM (and later, under MBM). Furthermore, the decline in the low-income rate under LICO, the declines in the low-income gap ratio and severity index from 2007 to 2009 were all marginal. As such, the trend in low income among lone parents was different from that of the overall population.
The evolution of low income for lone parents from 1976 to 2009 seemed to coincide with an increasing supply of labour by lone parents. Figure 3.5 shows the percentage of lone parents working 50 weeks or more in this period. Before 1998, the percentage of lone parents working 50 weeks or more varied, but never exceeded 40%. Since then, a strong upward trend emerged and the percentage of lone parents working 50 weeks or more usually exceeded 50%.
Figure 3.5 Percentage of lone parents working 50 to 53 weeks per year
Nevertheless, the change in low income for lone parents from 2007 to 2009 leads to an interesting question: is low income among lone parents less sensitive to a recession than that for the population as a whole? One direction for further investigation would be to look at the industries or occupations in which lone parents worked. For example, from 2007 and 2008, the GDP of goods-producing industries declined by close to 4%, while that of the service sector increased 5%. The vast majority of lone parents are women, and given their increased labour force participation one would not expect an increase in the low-income rate for them if they tended to work in industries or occupations less sensitive to a recession.
Unattached non-elderly people
Here we define unattached non-elderly people as those aged 45 to 64 who live alone. Low income incidences for this group of people for the 1976-to-2009 period are contained in Figure 3.6. Low income incidence, as well as the gap ratio and severity for them, were generally high among this group of people. For example, the incidences remained above 30% under LICO, LIM and MBM, even during the 2000-to-2009 period. Likewise, their low income gap ratios and severity indexes (not shown) were also much higher than those for the general population.
There might be several factors behind the high low-income indexes among this group. First, the upper age limit of 64 suggests that they are not at normal retirement age and thus a significant proportion of them were not entitled to retirement pensions and other old age benefits. Second, they may suffer other risks and vulnerabilities. For example, from 2000 to 2009, about a quarter of the group also had disabilities.4 Finally, the way low income is measured implies that, other things being equal, low-income statistics for people living alone would be higher than those for people living with others, because they do not enjoy economies of scale in consumption as much as others.
Figure 3.6 Low-income rates for unattached non-elderly people and percentage of those people not working
Given relatively low labour force participation rates, it is tempting for one to link government transfers to this group of people with their low-income status. Further investigation shows that labour market activity, rather than government transfers, was strongly correlated with low income for unattached non-elderly people. A particular group characteristic was that a significant portion of them were not working. Figure 3.6 (right axis) indicates that, from 1976 to 1980, about one-third of them did not work. This increased steadily to peak at 45% in 1998. Since then, the percentage of unattached people not working decreased to 35% by the 2007 to 2009 period. This seems to suggest that their low-income rate evolved in concert with the proportion of non-elderly unattached people not working.
In contrast, government transfers evolved differently from their non-transfer incomes. The average and median after-tax income of this group of people increased significantly over time. The only exception was during the 1990s, when little change occurred. On the other hand, government transfers to this group increased rapidly from 1976 to the early 1990s, but stayed at the same level thereafter.5 From 1976 to 1994, the ratio of average government transfer to average after-tax income for this group increased steadily from 10% to 23% and then declined rapidly to about 14% by 2007 and thereafter.
Comparing the evolutions of income and government transfers with the evolution of low income for the unattached non-elderly people, it can be seen that: (a) despite the increase in income from 1976 to the early 1990s, low income among the group did not decline, and (b) despite stable government transfers after the early 1990s, low income still declined. Hence, the relationship between government transfers and low income for this group was mixed, while the relationship between their labour supply and low income was fairly strong over the long term6.
It is well known that the earnings of recent cohorts of immigrants deteriorated over the past two decades. Moreover, there is also evidence suggesting that the labour market participation rate and the employment rate of new immigrants have deteriorated.7 Other things being equal, these factors imply that new immigrants would have a higher chance of falling into low income than their counterparts in previous decades.8 But the evolution of low-income among new immigrants was actually more complex than this. Figure 3.7 presents the low-income rates for new immigrants for the 1976 to 2009 period.9 The figure shows that the low-income rate among new immigrants was relatively low in the 1970s. It had tripled by the mid 1990s and declined substantially since then. In more recent years, however, the low-income rate for new immigrants started an upward trend.
The deterioration of new immigrants' earning capacity was associated with a greater gap between the earnings of new immigrants and the Canadian-born. This is directly reflected in the low-income statistics for new immigrants. But other factors also played an important role. In particular, the evolution of low income among new immigrants appears to be strongly affected by the total employment of the families of new immigrants. Figure 3.7 contains the average number of weeks worked by family members of new immigrants for the 1976-to-2009 period. It had declined from the 1970s to the lowest level in the mid-1990s. During the 1990s, it varied around 30 weeks per year. As a result, their income stagnated for most of the decade. But starting at the end of 1990s, an upward trend of employment for new immigrants began, matched by declines in the low-income rate and other indicators.
The increase in the low-income rate among new immigrants and the fact that it was higher after the 1990s might also be affected by a less well-known factor: the proportion of students among new immigrants gradually increased over the past 30 years. Before the mid 1990s, about 15% of new immigrants were students. This was similar to the proportion of students among the Canadian-born and older immigrants. But after the mid 1990s, the proportion of students increased to more than 25% among new immigrants, well above the 18% for older immigrants and those born in Canada.10
Figure 3.7 Low-income rates among recent immigrants and their weeks worked per year
As discussed in the previous chapter, low income among the general population continued a declining trend up to 2007. A similar trend was found among new immigrants when using the MBM.11 However new immigrants' low-income rate, when measured under LICO and LIM, increased slightly from 2005 to 2007. We also examined the higher order low-income indices and found that changes in the gap ratio and severity indexes under all three lines followed the same declining pattern in low-income rate as that under the MBM. This evidence suggests that low income statistics under LICO and LIM respond differently to those under MBM for new immigrants. As noted in Zhang (2010), new immigrants are most likely to settle in large cities where costs of living are typically high. At the same time we know that the same LICO and LIM threshold apply to all large cities whereas the MBM is different in each city12.
Figure 3.7 also indicates that the onset of the recession in 2008 affected the low-income rate for new immigrants. As their weeks worked in 2008 dropped abruptly, their low-income rates increased by 3 percentage points under LICO and about 4 percentage points under both LIM and MBM compared with those in 2007. However, only the change under the MBM was marginally significant. The gap ratio and severity indexes under different lines also increased, but these changes were not significant.
Off-reserve Aboriginal people
Low income among Aboriginal people is a concern in Canada. However, the SLID data only allows us to examine low income among off-reserve Aboriginal people for the period from 1996 to 2009. We identify a person as Aboriginal if he or she meets any one of the following three criteria: (1) a person reported he or she is a member of the Aboriginal target group for Employment Equity purposes; (2) a person reported having an Aboriginal background; or (3) a person reported himself or herself as a Treaty Indian or Registered Indian as defined by the Indian Act of Canada.
Figure 3.8 contains the trends in the low-income rate and gap ratio for off-reserve Aboriginal people for the period from 1996 to 2009 under LICO, LIM and MBM. Like the low-income rate for lone parents and new immigrants, the low income rate for off-reserve Aboriginal people dropped under all three lines in recent years. The decline in the low-income rate was matched closely by declines of the gap ratio (Figure 3.8, bottom panel) and the severity indexes (not shown). However, the low-income rate under LIM showed a flatter trend since 2000, although this was not accompanied by similar changes in the corresponding gap ratio and severity indexes. Indeed, the higher order low-income statistics under LIM increased briefly from 2000 to 2003 and, as with the other two lines, showed a declining trend from 2004 to 2008. Statistical tests suggest that the gap ratio and severity indexes under the three lines all declined significantly from 2003 to 2008.
As with other groups of people, low income among off-reserve Aboriginal people is strongly tied to their employment patterns. The upper panel of Figure 3.8 indicates the proportion of Aboriginal people (18 to 64 years old) who worked full-year full-time (right axis). From 1996 to 2000, the proportion increased by more than ten percentage points and stayed at about 46% thereafter (up to 2008). The low-income rate mirrored theses changes, but in the opposite direction: low-income rates under LIM and LICO declined about ten percentage points from 1996 to 2000.
Figure 3.8 Low-income rates for off-reserve Aboriginal people and percentage of them working full-year full-time (top) and gap ratios for off-reserve Aboriginal people (bottom)
From 2000 to 2008, the proportion of off-reserve Aboriginal people working full-year full-time was very stable (with the exception of 2002, in which the proportion reached a high point of 50%). Likewise, the low-income rate measured by the LIM was also very stable. However, this stable low-income rate under LIM was accompanied by a declining trend in the gap ratio and severity indexes. This was probably related to the fact that, during that period, the proportion of part-time workers among this group increased by several percentage points, which, while not necessarily reducing the incidence of low income, would help to reduce the gap ratio and severity index.
However, the recent recession seems to have affected off-reserve Aboriginal people. From 2008 to 2009, the percentage of this group of people (aged 18 to 64) working full-year full-time dropped by almost four percentage points to about 42%, the lowest level since 2000. This was immediately reflected in the increases of low income incidence and gap ratio for off-reserve Aboriginal people under all three low income lines.
People with activity limitations
The classification of people with activity limitations went through some revisions in the late 1990s. We focused on years since 1999, in which a person is identified as having activity limitations if he or she reported having difficulty with daily living activities or if the person reported a physical or mental condition or health problem that reduced the amount or kind of activities that he or she could do at home, school or work.
We would like to point out that low income statistics under the current practices in Canada may well under-estimate the hardship experienced by people with activity limitations. For any person dealing with health issues , it is reasonable to assume that the cost of living would be higher than for an otherwise identical person. But none of the three low-income lines takes this into consideration. Indeed, low-income statistics for people with activity limitations are calculated in the same way as those for anyone else, and the statistics below are no exception. Nevertheless, we are paying particular attention to trends in low income rather than the level of the statistics.
Figure 3.9 presents low-income rates and gap ratio indexes for people with activity limitations.13 Low income under different lines appeared to have moved in different directions in the last decade: the low-income rate, gap ratio and severity indexes under LIM increased over time, while those under LICO and MBM decreased. For example, from 1999 to 2009, the low-income rate under LIM increased from 17% to 20%, while it decreased from 18% to 14% under LICO. But under MBM, there was little change from 2000 to 2009.
In 2004, even though low-income rates did not change much relative to the previous year under all lines, there were one-time abrupt increases in the gap ratio and severity indexes. For example, under both MBM and LIM, the gap ratios increased by about half a percentage point, and some of these changes were statistically significant. This probably reflects a much slower growth in income of people with activity limitations relative to the change in income for the whole population—our data shows that, from 2003 to 2004, the median income for the general population increased by about 3%, while the median income for people with activity limitations stayed constant.
Figure 3.9 Low-income rates (top) and gap ratios (bottom) for people with activity limitations, 1999 to 2009
Like low income for the general population, low income for activity-limited people increased in 2008 relative to those in 2007. The increase was strong under LIM, measured by the significant increases in the low-income rate and gap ratio. Interestingly, from 2008 to 2009, low income statistics did not change much for people with activity limitations.
In this chapter, we examined low income among groups of people who are often referred to as groups at risk of social exclusion. The result shows that, in the last few decades, low income measures improved the most for senior Canadians. Lone parents also experienced remarkable progress, but low income among them was still high relative to other groups. However, the situation among unattached non-elderly people was probably the most serious among all groups, even though there were improvements over the last ten years for them.
Furthermore, although the Canadian economy experienced almost uninterrupted growth over the last decade and many vulnerable groups benefited from that with significant declines in low income, not all of them benefited equally from the growth. For example, low income among senior Canadians and Canadians with activity limitations increased recently, at least under the relative measures.
Finally, low income among off-reserve Aboriginal people, children, recent immigrants or even people with activity limitations was about half as those among unattached non-elderly people or lone parents. There still may be pockets of people falling behind in other dimensions. In the next chapter, we shall focus on the regional dimension.
1. Children are defined as those persons 17 years old or younger.
2. The 95% confidence interval estimates in 2007 and 2008 were [0.0874, 0.1037] and [0.0823, 0.0994]. The two intervals overlap considerably.
3. We also examined low income for lone parents using the MBM line. The result under this line is essentially the same as that under LICO.
4. Low income among people with activity limitations was also high. This will be discussed later.
5. There was, however, some difference between men and women within the group. The average transfer for men decreased in this period, while that for women started to increase again in the late 1990s.
6. While a fairly strong relationship exists, between 2008 and 2009 unemployment and the number of EI recipients rose yet the low income incidence fell for the unattached non-elderly.
7. See, for example, Aydemir and Skuterud (2004) and references therein.
8. In this report new immigrants are defined as those who had lived in Canada from 2 to 10 years in the year they were observed.
9. The gap ratio and severity indexes evolved in a similar pattern as the low-income rate for virtually all years except in 2006 and 2007 under MBM. The difference shall be discussed later.
10. When students were excluded from new immigrants, low-income rate was generally reduced by 2 to 3 percentage points in recent years.
11. Under MBM, the decline in the low-income rate for the general population from 2005 to 2007 was significant at the 95% confidence level and that for new immigrants was only marginally significant.
12. The MBM in Toronto is 8.8% higher than the MBM in Montreal and 3.5% higher than the MBM in Vancouver. The proportion of immigrants in Toronto and Vancouver is at least twice the level in Montreal however Vancouver has a higher proportion of new immigrants than do Montreal or Toronto
13. The trend in the severity index is similar to that in the gap ratio.