Income Research Paper Series
Low Income in Canada - A Multi-line and Multi-index Perspective
Low income across provinces and cities
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 this chapter, we discuss the evolution of low income at sub-national levels and compare low income between provinces and cities. We also attempt to identify pockets of low-income people across provinces and cities, so as to provide context for place-based policy discussions. While the SLID sample allows us to examine low income for groups of individuals in practically all provinces, it does not allow us to do this across all cities. We focus on cities with a population of 1 million or more in 2009: Montréal, Ottawa–Gatineau, Toronto, Calgary, Edmonton and Vancouver, as well as Winnipeg.1
We examined the period from 1976 to 2009. More attention was paid to the 2000 to 2009 period, in which all three low-income lines were available and the definitions for several groups of individuals were consistent. Thus, when low income across groups of individuals within a province or city is discussed, the reference period is limited to 2000 to 2009. Low income rates in 2000 and 2009 for the provinces, selected cities and across various groups of individuals can be found from Tables 4.1 and 4.2 at the end of this chapter.2
The evolution of low income in provinces and cities
In Newfoundland and Labrador the incidence of low income dropped significantly, with particular strong declines from 1976 to 1989 and from 2000 to 2008. The decline from 1976 to 1989 resulted in the provincial incidence of low income approaching the national average. Similarly, from 2000 to 2009, low-income rates measured by all three lines declined. In 2000, the provincial low-income rates were 13.2% under the low-income cut-off (LICO), 21.4% under the low-income measure (LIM) and 20.5% under the market basket measure (MBM), and under both LIM and MBM, the provincial incidences were significantly higher than the corresponding national incidences in 2000.3 But in more recent years, the low-income rate under LICO was lower than the national rate, the differences between provincial and national rates under LIM and MBM were much smaller than in the early 2000s, and these declines in provincial low-income rates were accompanied by similar changes in the low-income gap ratios. However, under LIM and MBM, incidences of low income in this province were still significantly higher than the corresponding national level as of 2009.
Prince Edward Island also saw improvement. Its distinguishing characteristic was a persistent long-term decline in the incidence of low income under all three thresholds. Under LICO, the provincial incidence of low income was similar to the national incidence in the mid-1970s. Starting in the early 1980s, the incidence at the provincial level was consistently below the national level, and the differences were statistically significant in almost every year. The provincial incidence of low income under LIM was above the national level before the 1990s, but thereafter, the incidence at the provincial level was not statistically different from the national level.4 Likewise, under MBM, the incidence at the provincial level was higher than that at the national level in the early 2000s, but more recently the incidence at the provincial level matched that at the national level.
Nova Scotia followed the national trend for low income at the provincial level under all thresholds. Under LICO, the incidence at the provincial level was tied to that at the national level. But under LIM (from 1978 to 2009) and MBM (from 2000 and 2009), low-income rates in Nova Scotia were significantly higher than the national averages. A somewhat discouraging sign for this province was that, from 2006 to 2009, the gaps between the provincial and national incidences of low income under LIM and MBM became wider than they were before 2006.
New Brunswick had, in the mid-1980s, a higher incidence of low income under LICO than the national level. Differences between the incidence at the provincial and national level were almost always statistically significant. However, starting in the late 1990s the incidence of low income under LICO dropped below the national level, and continuing improvements were observed during the most recent years. As in Nova Scotia, the incidences under LIM and MBM in New Brunswick were higher than the corresponding incidences at the national level. But an encouraging sign was that low income improved under both LICO and MBM from 2000 to 2009, and there were also continuing improvements in the four years up to 2009 under LIM. Indeed, by 2009, the provincial low income rates under LIM and MBM were no longer significantly different from the corresponding national rates.
Quebec generally followed the national trend in low income from the early 1980's to the late 1990's. During this time the provincial low-income rates under LICO and LIM were generally higher than the corresponding national rate. But from the late 1990s to the early 2000s the incidence dropped significantly and dropped faster than that at the national level under LICO, such that by the late 2000s, the national and provincial incidences were no longer different from each other. Under LIM, the provincial low-income rate had also been higher than the national rate before the early 2000s. But thereafter, the provincial rate declined and started to approach the national rate. The early 2000s also witnessed significant declines in the incidence of low income under MBM for Quebec. But that trend reversed by 2005 when the incidences of low income started to inch up to the national level.
Ontario's low income rates, like Quebec's, rose around recessions and fell between them. But different from Quebec, the low income situation in Ontario deteriorated over time, particularly under LICO and LIM. In 1976 the Ontario low income rates were just over 80% of the national rates and they dropped to a low point of about 75% of national rates by the mid-1980's. Afterward the incidence of low income at the provincial level rose fairly steadily to approach the national level. By 2009 the incidence in Ontario under the LICO exceeded that of Canada for the first time (10.1% vs. 9.6%). The year 2004 was a watermark for low income under MBM: from 2000 to 2003, the provincial incidence was below the national incidence, while from 2004, the provincial incidence rose to the national level. In addition, Ontario was different from all its eastward provincial neighbors in that its low-income incidences and gap ratios rose significantly from 2007 to 2009, under both LIM and MBM.
Manitoba has seen improvement since 1976, when the incidence of low income was about 30% higher than those at the national level under LICO and LIM. Manitoba gained ground and was roughly equal to the national level in 1983-1984 but then rose to about 20% above the National level in the late 1980's and early 1990's. By the mid-late 1990s, the incidence of low income under LICO for Manitoba was similar to the national level. The incidence under LIM at the provincial level was higher than that at the national level, and often significantly. Both the incidence of low income and the gap ratio have stayed relatively flat since the early 1980s in Manitoba, and with the national incidence following an upward trend since the late 1980s, the provincial and national difference became smaller in more recent years.5 More remarkably, the low-income gap ratio and the severity index dropped in Manitoba between 2000 and 2009 under all three lines. Another positive sign was that the provincial incidence under MBM was consistently below the national level, sometimes significantly.
The low income rate under LICO passed a watershed in Saskatchewan in the mid-1990s. Before that, the incidences of low income at the provincial and national level were similar; afterwards the provincial-level incidence dropped below the national, and the difference was often statistically significant. The incidences under LIM and MBM for the province were consistently higher than the corresponding national ones, and the differences under LIM were particularly large and significant before 2009. But encouragingly, for all lines in years leading up to 2008, when the incidence at the national level dropped, the incidence at the provincial level dropped faster, and when the incidence at the national level increased or stayed flat, the incidence at the provincial level continued to decline. Hence, it is not surprising to observe that by 2009, the provincial low-income rates under both LIM and MBM dropped to a historical low, dropping below the national level for the first time.
Alberta saw incidences of low income consistently below the national level under all three thresholds, except during the decade from the mid-1980s to the mid-1990s. A strong downward trend in the incidence started in the early 1990s, leading to significantly lower incidences under all three thresholds for Alberta, compared to the national level, from 2003 to 2008. However, the province was hit hard by the recent recession. In 2009, low-income incidences, gap ratios and the severity indexes under all three lines became significantly higher than those in 2008, and the increase was particularly strong under MBM with the incidence rising four percentage points to 10% between 2008 and 2009.
British Columbia generally followed the national low-income trend. However, when low income increased at the national level, the provincial low income increased faster and when the national low income decreased, the provincial low income decreased faster. From 1976 to 1983 the incidence of low income was lower in BC than in Canada. From there until 1998 the rates were quite similar under both LICO and LIM. However, from the late 1990s onward, the provincial incidence started to surpass the national level under the two lines, and the difference was often statistically significant. As well, under MBM, low-income incidence in this province had always been above the national level.
Montreal's incidence of low income stayed relatively flat from 1976 to the early 1990s under both LICO and LIM, while during the rest of the 1990s, it increased. But over the next ten years, Montréal residents enjoyed significant reductions in the incidence of low income. Nevertheless, low-income incidence in Montréal was still one of the highest among large cities in recent years, and in the last two years, there was evidence to suggest that low income in Montreal increased again. For example, under MBM, the low-income incidence in 2008-2009 was significantly higher than in the 2004-to-2007 period. The incidence under LIM also increased, and the increase in 2008 was particularly strong.6
Ottawa–Gatineau saw highs and lows in the incidence of low income in the 1976-to-2009 period. As in Montréal, Ottawa–Gatineau's incidence of low income reached a historic peak in 1996 and 1997. Over the following ten years, residents of the national capital region experienced declining incidences of low income. But in 2008, low-income incidences under all three lines increased compared to the previous year. The gap ratio under LIM also rose significantly. However, the overall declining trend had not been altered as low income dropped again in 2009 relative to that in 2008 (although not significantly).
Toronto experienced some significant reductions in low income under both LICO and LIM during the late 1980s. But as in other large cities, the low-income incidence under LICO varied dramatically in Toronto, ranging from less than 8% in 1988 to more than 17% in 1995. During the last ten years, from 2000 to 2009, while the low-income incidence under LICO and MBM fluctuated between 10% and 13%, the incidence under LIM followed a strong upward trend, increasing from about 9% to 13%, and a similar trend was observed in the low-income gap ratio and the severity index under LIM.7
Winnipeg residents had a low-income rate in 2008 similar to that in 1976 under LIM. As well, no clear trend in the LICO rate can be seen before the 1990s. But from the mid 1990s to 2009, the low-income rate under the LICO declined significantly. Furthermore, the low-income gap ratio and severity index declined under all three lines during the 2000-to-2009 period, although only the drops in the LICO gap ratio and severity index were significant.
Calgary's incidence of low income under both LICO and LIM increased steadily from 1976 to the early 1990s. Thereafter, the incidence dropped back to the mid-1970s level. However, the reduction in low income from the early 1990s to 2008 was dramatic. For example, at the peak around 1992, the LICO and LIM incidences in Calgary were practically the highest among all major cities. By 2000, low-income rates for this city fell to the lowest among all large cities: 10.0% under LICO; 7.0%, LIM; and 8.9%, MBM. By 2008, they were still the lowest: 6.6% under LICO; 5.1%, LIM; and 6.0%, MBM. However, the latest recession led the LIM gap ratio to double over the level in 2007.
Edmonton, like Calgary, saw rising low-income rates from 1976 to the early 1990s and dramatically falling rates thereafter. In contrast to Calgary, low-income rates under LICO and LIM in this city fell somewhat below their mid-1970s levels. Reductions in low income from the early 1990s to 2008 were also strong in Edmonton. In 2000, the incidences were 13.7% under the LICO, 10.1% under the LIM; and 10.1% under the MBM. By 2008, these dropped significantly to 7.0% under the LICO; 7.1% under the LIM, and 4.9% under the MBM. But the latest recession struck Edmonton hard.8 The incidence, gap ratio and severity indexes under all three lines increased in 2009 relative to the 2006-2008 level. For example, the incidence under MBM was about 5% in the three years leading to 2009. It reached 12% in 2009.
Vancouver's low-income situation was mixed. From 1976 to the early 1990s, low-income rates under both LICO and LIM increased much of the time. The city's incidence of low income under LICO declined from the late 1990s to 2007. This was accompanied by the declining incidence under MBM from 2000 to 2007. But the incidence of low income under LIM followed a strong upward trend starting from the late 1980s. However, the gap ratio and severity indices dropped uniformly under all three lines in the period from 2000-to-2007. The latest recession also struck Vancouver, resulting in greater low-income incidence, gap ratio and severity in 2009 than in 2007 under all three thresholds.
Comparisons of low income among provinces and cities
A discussion of strategy and rules of comparison
Robust comparisons of low income between provinces or cities are not straightforward, nor are they free of controversy. Our approach is to rely on multiple low income lines and multiple indexes. We also take into account the evolution of low income for a province or a city.
Using multiple lines is desirable because low income can be measured from different angles. A region might have done very well under a relative line, but under an absolute line, a different picture may appear. Hence a comparison with a single line is not as robust as that with multiple lines. Employing multiple indexes is desirable because different indexes have different properties and they provide different information. A comparison using a single index can be misleading. For example, a region may have high incidence, while its gap ratio might be quite low, or vice versa.
It is also desirable to compare low-income at different points of time. A comparison at any particular point in time may not be robust because regional economies may evolve along different paths. In a particular year, some regions might be already out of a recession while others might still be in a trough. Of course, it was not practical to make comparisons at every point of time. We report on our low-income ranking for 2009, but the result is not based on the 2009 low-income statistics only. The evolution of low income from 1976 to 2009 (from 2000 to 2009 under MBM) in a province or city is also taken into consideration.
We employed five rules to help us compare low income between provinces and cities. First, we based all comparisons on rigorous statistical tests. Low-income indices themselves are estimates from survey samples, and are thus affected by sampling variations. Two provinces or cities can only be meaningfully compared if differences in their low-income indices are statistically significant.9 In particular, if the estimated confidence interval of a low-income index for one province overlapped that of another province, then the two provinces were inferred to have the same level of low income. If the two intervals did not overlap, or if they overlapped only slightly, we would say that the two provinces were significantly or marginally different from each other, respectively.
Second, while we focus on low-income incidence estimates for the comparisons, the gap ratio and the severity index are also employed as complementary factors. The gap ratio is a higher-order index than the incidence; the severity index is a higher order index than the gap ratio. Hence, if two provinces or cities had the same incidence, but one had a higher gap ratio, then the one with higher gap ratio is ranked below the other. Similarly, when both the incidences and the gap ratios are the same, low-income severity is employed in the comparison.
Third, when different thresholds gave different results in low-income comparisons, we put lower weight to results under the LICO than to those under MBM and LIM. In the literature, low-income lines such as the LIM are typically viewed as thresholds to help identify individuals who do not have the resources to fully participate in community life. In this sense, a person with income below the LIM threshold is likely to be at high risk of social exclusion. Since the LIM thresholds are rebased every year, they would reflect the contemporary standard of living. On the contrary, the current LICO thresholds are based on Canadian households' 1992 standard of living. Although they have been updated according to changes in the consumer price index every year, their real values remain at the 1992 level.10
Fourth, when the rules described above could not help us make a comparison, we examined how low income had evolved within a province or city. If two regions were the same according to the rules above, we would use the relative position of a province or city to determine its rank. For example, if two provinces had the same incidence of low income, gap ratio and severity in 2009, but one had higher incidence in 1976 or 2000 than the other, it would be reasonable to say that the province with the higher incidence in 1976 was ahead of the other province because it progressed more in reducing low income over time. Finally, we conducted stochastic dominance tests of various orders to rank the provinces if the first four rules failed to identify the position of a province or city.
Of course, the above rules are not without caveats. On the one hand, even though they are reasonable to us, they will still be debatable to some readers. On the other hand, there is no guarantee that the comparisons are complete. For example, if the estimated confidence interval of a low income index for one province is so wide that all the intervals for other provinces fall between the upper and lower bounds of the estimated interval for the province in question, then the province cannot be ranked. In addition, all three low-income lines are based on certain assumptions and expert judgments. Our results will be subject to bias because of these assumptions and judgments. Nevertheless, we believe that the multiple-period, multiple-line, multiple-index approach is an important improvement over a single-point, single-line, single-index approach in low income comparisons. Our hope is that the simple, transparent rules we have created can provide a basis for further discussions.
Low-income comparisons between provinces
Prince Edward Island fared best among all of the ten provinces. In terms of low-income incidence, the province of Alberta was ahead of PEI under LIM in 2009. But under LICO, PEI was ahead of Alberta, while under MBM, the incidences of the two provinces were tied. However, when the gap ratios were examined, PEI was the best among all ten provinces under all three lines, and the severity indexes in PEI were practically the lowest over all other provinces in 2009. In addition, the evolution of low income in PEI was positive. It advanced from the middle of the country's low-income ladder in 1976 to the top by 2009.
The province of Alberta followed PEI closely. Indeed, if we were to use low-income incidence alone with a stronger emphasis on the LIM and MBM lines, Alberta would have been ranked ahead of PEI.11 But the low-income gap ratios and the severity indexes in Alberta were much higher than those of PEI. Indeed, the gap ratios and the severity indexes in Alberta were even higher than those in several other provinces, such as Saskatchewan, Manitoba and Newfoundland and Labrador. In addition, Alberta was in a relatively better position in 1976 in the low-income ladder and hence its long-term progress was not as strong as that of PEI.
Saskatchewan followed Alberta. In 2009, under LIM, the incidence in Saskatchewan was only higher than that of Alberta, basically the same as those in PEI and Ontario, and lower than those of the other six provinces. Under MBM, its incidence was similar to the two front-runners (PEI and Alberta), while under LICO, Saskatchewan's incidence was only higher than that of PEI. In addition, this province had made strong progress in low income over time. In 1976, it was ranked in the upper middle, while by 2009 it became one of the top three provinces according to the low-income incidence. We noticed that Saskatchewan's low-income gap ratio was similar to that of Manitoba and Newfoundland and Labrador, while its low income severity was higher than both provinces in 2009. But when stochastic dominance tests were conducted, Saskatchewan dominated Manitoba, Newfoundland and Labrador and Ontario in second order tests.12
The above comparison leaves Manitoba, Ontario and Newfoundland and Labrador to compete for the next spot. In terms of low-income incidence, the evidence was mixed. Ontario ranked higher than Manitoba and was similar to Newfoundland and Labrador under LIM. Under MBM, Manitoba ranked ahead of Newfoundland and Labrador and was similar to Ontario. But under LICO, Ontario's incidence was higher than that of Newfoundland and Labrador and the same as Manitoba. The gap ratio and severity indexes showed similar mixing patterns between these provinces. However, when we put more weight on incidences based on LIM and MBM than on those based on LICO, Ontario was ranked ahead of the other two provinces.
Newfoundland and Manitoba made more progress in low-income reduction over time than did Ontario from 1976 to 2009, but results from dominance tests suggest that Ontario dominated Newfoundland and Labrador in second order for a threshold up to $13,977 in 2009. Ontario also dominated Manitoba in second order for a threshold up to $18,846, while Manitoba dominated Newfoundland and Labrador in second order for a threshold between $1,002 and $12,245. Hence, we would rank Manitoba after Ontario and Newfoundland and Labrador after Manitoba in 2009.
Quebec and New Brunswick seemed to be in line for the next spot. It is hard to distinguish them in terms of their low-income incidence in 2009. Under LIM, the two provinces were similar. But under MBM, the incidence in Quebec was marginally higher than that in New Brunswick, while under LICO, the incidence in New Brunswick was significantly lower than that in Quebec. On the other hand, under MBM and LIM, the gap ratio and the severity indexes in Quebec were marginally lower than those in New Brunswick. However, New Brunswick made more progress over time than Quebec did under LIM and MBM. Those uncertainties were made a bit clear with dominance tests since we found that Quebec dominated New Brunswick in second order tests for any give threshold. Hence, we would rank Quebec ahead of New Brunswick.
The two remaining provinces, Nova Scotia and British Columbia, had similar incidence, gap ratio and severity index estimates under both LIM and MBM in 2009, but under LICO, the indexes in Nova Scotia were significantly lower than those in BC. In addition, under both LICO and LIM, Nova Scotia made more progress in low-income reduction in the period from 1976 to 2009, although under MBM and LIM, BC progressed more than Nova Scotia from 2000 to 2009. The overall evidence suggests that Nova Scotia was slightly ahead of British Columbia.
Low-income comparisons between major cities
Calgary and Ottawa-Gatineau shared top spot in having the least low-income among the major cities in 2009. Under LIM, the low-income incidence for Edmonton was not significantly different from those of Calgary and Ottawa-Gatineau. Under the MBM, Winnipeg was also a potential candidate, while under the LICO, Winnipeg and Edmonton were the competitors. But since Edmonton had marginally higher incidence than Calgary and Winnipeg, and significantly higher gap ratio and severity than Winnipeg, we excluded Edmonton from the candidates for the top spot.
In 2009 Winnipeg had marginally lower gap ratio and severity than Calgary under MBM and its historical progress was impressive under both LICO and MBM. But its incidence under LIM was significantly higher than that in Calgary. Results from dominance tests also show that Calgary and Ottawa-Gatineau dominated Winnipeg13. This leaves Calgary and Ottawa-Gatineau for the comparison. In terms of the three low-income lines and indexes, the two cities were not statistically different from each other. But Ottawa-Gatineau made stronger progress than Calgary from 1976 to 2009, while results from stochastic dominance tests suggest that Calgary dominated Ottawa-Gatineau. Hence, we failed to distinguish the two cities.
Between Edmonton and Winnipeg, Winnipeg was ahead of Edmonton. Under both LICO and LIM, these two cities were not different in low-income incidences, gap ratios and the severity indexes in 2009. But under MBM, the incidence and the severity index in Winnipeg were marginally lower than those in Edmonton, and the gap ratio of the former was significantly lower than in the latter, largely due to a significant increase in low income in Edmonton in 2009. In addition, low-income incidence under LICO dropped strongly between 1976 and 2009 in Winnipeg, from 19% to 10%, suggesting Winnipeg enjoyed a strong historical progress in low income reduction.
Among the rest of the three cities, Vancouver had a clear disadvantage in low-income comparisons under all lines and all indexes. It is thus a matter of ranking the other two cities: Montreal and Toronto. In 2009, these two cities had no significant differences under the three low income lines and indexes. However, low income had been stronger in Montreal than in Toronto historically. For example, low-income incidence in Montreal in 1976 was about 18%. By 2009, it dropped to about 13%. But results from dominance tests suggest that Toronto dominated Montreal in second order for a threshold up to $14,469. As a result, we put Montreal and Toronto in the same position on the low-income ladder.
The disadvantaged position of Vancouver's low income can be clearly observed under all three low income lines and indexes. Under MBM and LICO, the incidences, gap ratios and severity indexes in Vancouver were significantly higher than those in other cities. While under LIM, Vancouver's low-income incidence was significantly higher than those in five other cities, only Montreal had a similar incidence. But the gap ratio in Vancouver was higher than all cities, including Montreal. Hence, we put Vancouver last among the seven major cities.
Low-income groups across provinces and cities
Low income groups across provinces
Newfoundland and Labrador saw significant declines in low-income rates for children, people with activity limitations and to a less extent, lone parents, from 2000 to 2009, under all three low-income lines. But compared to the national level, the province still had pockets of individuals facing economic hardship in 2009. For example, under both LIM and MBM, low-income rates for lone parents and children were much higher than the national average.
Prince Edward Island saw little gain in the reduction of low-income incidence from 2000 to 2009 among children, seniors, lone parents, and people with activity limitations.14 Under all three lines, the 95% confidence interval estimates of low-income rates in 2009 were not statistically different from those in 2000 for these groups of individuals. In addition, low-income rates for lone parents were still significantly higher than the national averages under both LIM and MBM in 2009. Hence, the observed improvement in low income in this province did not benefit any of these groups of individuals.
Nova Scotia also had small reductions in low-income across several vulnerable groups from 2000 to 2009. At the same time, low-income rates under LIM and MBM for seniors, lone parents and off-reserve Aboriginal persons in 2009 were all significantly higher than the corresponding national averages.
New Brunswick's changes in low-income rates between 2000 and 2009 for the vulnerable groups (children, seniors, lone parents and people with activity limitations) were not statistically significant under any threshold. But low income rate among unattached non-elderly persons did drop marginally under both LICO and MBM. When compared to national averages, however, we observe that lone parents living in this province still had much higher incidence under both LIM and MBM (40 and 36%, respectively) in 2009.
Quebec made significant progress in low income reduction for children and lone parents between 2000 and 2009. In 2000, the low-income rate for children was 16.1% under LICO; 17.9% under LIM; and 13.7% under MBM. Nine years later, this dropped to 7.7% under LICO; 12.8% under LIM; and 8.9% under MBM. While in 2000, low-income rates for lone parents were 37.9% under LICO; 40.3% under LIM; and 36.0% under MBM. By 2009, they dropped to 18.2% under LICO; 28.4% under LIM; and 21.0% under MBM, respectively. Low income among seniors in Quebec was still higher than the national average in 2009 under both LICO and LIM.
Ontario saw an important improvement for lone parents in low income from 2000 to 2009. Under both LICO and MBM, low-income rates for lone parents dropped significantly. In 2000, they were 29.0% (under LICO) and 28.3% (under MBM). By 2009, these dropped to 18.2% and 20.5%, respectively. But low-income rates for people with activity limitations increased under LIM and MBM, while remained largely the same under LICO. For this group of people, the low-income rate increased by seven percentage points under LIM, from 13.1% to 20.1% between 2000 and 2009, while under MBM, it increased nearly five percentage points, from 11.2% to 14.7%. However, in Ontario, none of the vulnerable groups we examined had significantly higher incidence than their counterparts in other parts of the country in 2009.
Manitoba showed progress in reductions in the low income of lone parents during the 2000-to-2009 period. Under both LICO and MBM, low-income rates for lone parents from this province dropped. In 2000, their low-income rates were 42.2% and 40.7% under LICO and MBM. In 2009, these declined to 21.9% and 24.2%, respectively. However, if a relative LIM is employed, the low-income rate for this group of individuals did not change much between 2000 and 2009 and in 2009, their low-income rate was still significantly higher than the national average.
Saskatchewan's overall improvement in low income reduction between 2000 and 2009 did not equally benefit the various vulnerable groups. But children saw some progress. In 2000, low-income rates for children were 16.9% and 21.5% under LICO and LIM. These dropped to 9.1% and 15.2%, respectively by 2009. More importantly, low income rates for unattached non-elderly individuals from this province dropped significantly under all three lines. In 2000, around 50% of them were in low income, but by 2009, only about one quarter of them was in low income. Furthermore, people with activity limitations did well when compared with their counterparts from the rest of the country in 2009 under all three low-income lines, while in contrast, lone parents from this province had significantly higher incidence when compared with lone parents from the rest of the country under any threshold in 2009.
Alberta's improvements in its low-income rates occurred for all groups of at-risk people by practically each measurement from 2000 to 2008. But the latest recession reversed much of the gains. Only off-reserve Aboriginal people and people with activity limitations saw marginal improvements between 2000 and 2009 under LICO and MBM. However, when compared with the rest of the country, seniors, off-reserve Aboriginal people and people with activity limitations all had significantly lower incidences in 2009, no matter which low-income line was employed for the comparison.
During the 2000-to-2009 period British Columbia posted low-income rate reductions across two groups of at-risk people under both LICO and MBM: lone parents and people with activity limitations. Under LICO, their low-income rates dropped from 28.7% to 14.8% and from 20.6% to 14.3%, respectively, while under MBM, the corresponding changes were: from 34.8% to 18.1% and from 20.7% to 14.8%. However in 2009, none of the at-risk groups of individuals had significantly different low-income rates than their counterparts in the rest of the country.
Low income groups across cities
Montréal saw remarkable improvements for lone parents in the nine-year period from 2000 to 2009. In 2000, low-income rates for lone parents were 42.8% under LICO; 42.9%, LIM; and 36.2%, MBM. These were practically the highest among lone parents living in large cities. But by 2009, the corresponding low-income rates dropped to 23.6% under LICO; 28.9% under LIM; and 22.0% under MBM. Children and people with activity limitations living in Montréal also saw declining low-income rates under LICO, but not under the two other lines. However, the low-income rate for seniors increased under the MBM. When compared with the averages of the seven large cities, none of the vulnerable groups in Montreal had a higher incidence in 2009 under any of the three low-income lines.15
Ottawa–Gatineau posted no significant changes in low income among the several vulnerable groups we examined and their low-income rates were not different from the corresponding city averages over the 2000 to 2009 period. The only group that had significant (under LICO) or marginally significant (under LIM and MBM) lower incidences were seniors.
Toronto's positive development during the 2000-to-2009 period was the declining low-income rate among lone parents. In 2000, the low-income rate for lone parents in Toronto was 30.3% under LICO and 28.7% under MBM. By 2009, both had dropped to 14.5%. But the low-income rate among people with activity limitations increased under both LIM and MBM. In 2000, their incidence was 9.5% and 10.0% respectively, increasing to 20.1% and 18.7% in 2009.
Winnipeg saw improvements in low income among children, lone parents, off-reserve Aboriginal people and people with activity limitations, but only under the LICO. Under the other two thresholds, not much change was observed during the 2000-to-2009 period. On the other hand, the low-income rate for children in 2009 was significantly lower than the city average, while the low-income rate for lone parents was significantly higher than the city average.
Calgary and Edmonton's relatively small sample sizes for various vulnerable groups did not allow us to make many comparisons. The only group we could identify as having improved their low-income rate during the 2000 to 2009 period was people with activity limitations from Edmonton, for whom the incidence dropped from 18.2% to 8.9% under the LICO. But compared to the city averages, several groups did well in 2009 in these two cities. In Calgary, seniors, children and people with activity limitations all saw lower incidences than the city averages under the different low income thresholds. On the other hand, in Edmonton, while both seniors and people with activity limitations had a lower incidence in 2009 under the three thresholds, the low-income incidence for children under MBM was significantly higher than the city average.
Vancouver's lone parents experienced some reduction in the low-income rate under the LICO. In 2000, their low-income rate was 30.2% and by 2009 this rate had dropped to 9.4%. There were some drops under LIM and MBM, but these were not statistically significant due to small sample sizes.16 The incidence of low-income was higher among seniors in 2009 under LIM and MBM, although these increases were only marginal. On the other hand, the incidence of low income for vulnerable groups was not much different than that of the general Vancouver population under all low-income lines in 2009.
The low-income rates evolved in different ways across the provinces and cities. Some provinces and cities benefited a great deal from the overall economic growth of the decades leading up to 2008. Significant progress was made in Alberta, Prince Edward Island, and Newfoundland and Labrador, as well as in Edmonton, Calgary, and Winnipeg. At the same time, improvements were seen with respect to low income for children and lone parents in most provinces and cities. However, the recent recession had an important impact on low income in several provinces and cities, most notably Alberta and its capital city Edmonton.
1. Winnipeg had slightly more than 742,000 residents in 2009. It was selected because the number of people sampled from that city in Survey of Labour and Income Dynamics 2009 was fairly large (2,464 observations).
2. Tables A4 and A5 in the Appendix contain the 95% confidence intervals for Tables 4.1 and 4.2 respectively
3. Because the provincial and the national samples are not independent, the comparison was actually made between the provincial sample and the sample consisting of those from the other nine provinces. In what follows, whenever we make a test between a region and the nation, or a group from a region with a group at the national level, the national sample represents those from all other regions, with the region in question being excluded ,so that the two samples are not overlapping.
4. Although from 1999 to 2002, the provincial level was above the national level.
5. Although in 2009, the provincial incidence under LIM increased and became significantly higher than the national level while under the LICO the incidence dropped to 93% of the national level in 2008 and remained there in 2009.
6. There was a decline in low-income incidence in 2009 under LICO, but this was not supported by the corresponding drops in the gap and severity indexes.
7. The upward trend in the gap ratio and severity indices had started in the late 1990s.
8. Only Vancouver had a similar situation as in Edmonton.
9. All of the comparisons to follow are based on comparing the 95% confidence intervals on the corresponding low income indices. This practice has been followed by Osberg and Xu (1999), among others.
10. In terms of the ability to reflect the changing standard of living, the MBM threshold falls between the thresholds of LICO and LIM because it is rebased periodically.
11. As noted before, the latest recession had a large effect in Alberta's low income in 2009. If we were to rank the provinces according to low income in 2008, Alberta would have been ahead of PEI.
12. Saskatchewan dominated Manitoba, Newfoundland and Labrador, and Ontario at the second order, when the threshold was set between 0 and $18,785, between 0 and $15,750 and between $8088 and $18,692, respectively.
13. Calgary dominated Winnipeg in first order tests for a threshold between $2,745 and $13,821, while Ottawa-Gatineau dominated Winnipeg in second order tests for a threshold up to $17,742.
14. The numbers of observations for other groups such as recent immigrants or off-reserve Aboriginal people were not large enough to make meaningful inference on their low income situation.
15. New immigrants had fewer than 100 observations in Montreal and were thus not examined.
16. There were only 103 observations in 2009 for lone parents living in Vancouver.