Garnett Picot, Statistics Canada
John Myles, University of Toronto and Statistics Canada
Garnett Picot is Director General of the Socio-Economic and Business Analysis Branch at Statistics Canada.
John Myles is a Professor in the Department of Sociology at the University of Toronto, and a Research Fellow at Statistics Canada.
Trends in low-income levels and income inequality in Canada are two of the more closely watched indicators of economic well-being. In this paper we review recent evidence regarding these trends in Canada.
Family Income Inequality in Comparative Perspective
For cross-national comparative purposes, our selection of countries is based on a well-established body of comparative research. Reflecting long-standing institutional and political differences, these clusters include the Scandinavian/Nordic nations, represented here by Finland and Sweden, the countries of Continental Europe (Belgium, Germany, and the Netherlands) and the predominately English-speaking countries (Canada, the United Kingdom, and the United States).
Table 1 summarizes recent (1997-2000) results on cross-national differences in income inequality for these eight nations. Measured with the Gini index (the higher the Gini, the greater the inequality), family income inequality was remarkably similar among the Central European and Nordic nations by the end of the century. Canada’s position (Gini = .29) as more egalitarian than the United States (Gini = .37), and the United Kingdom (Gini = .35), but less equal than the countries of Central and Northern Europe, replicates a familiar pattern found in many earlier cross-national comparisons.
|After-Tax/Transfer (Disposable) Income Adult Equivalent Adjustedb|
|Gini||Ratio of High to Low Income (P90/P10)||Ratio of Low to Middle Income (P10/P50)||Ratio of High to Middle Income (P90/P50)|
|United States (2000)||0.37||5.43||0.39||2.10|
|United Kingdom (1999)||0.35||4.54||0.47||2.14|
|The Netherlands (1999)||0.25||3.27||0.53||1.75|
a The Gini coefficient varies between 0 and 1.0: the higher the Gini, the greater the level of inequality. The results in Table 1 are based on family disposable (after-tax and transfer) income, where the unit of analysis is the individual, and each individual is attributed the family “adultequivalent adjusted” family income.
b Adult equivalent adjusted income is a per capita measure of income that accounts for the economies of scale associated with larger families. It is computed by dividing family income by the number of “equivalent adults” in the family. A family of four may have two equivalent adults, based on the assumption that four people in one household is the equivalent of two people living on their own.
Source: Smeeding (2003).
|Late 1970s||Mid-1980s||Mid-1990s||Most Recent|
Source: LIS key figures.
Comparing percentile ratios provides a more intuitive understanding of these differences. In the United States, family incomes near the top of the distribution (the 90th percentile) are over five times higher than family incomes near the bottom of the distribution (the 10th percentile). In Canada, the ratio is about four to one, while in Sweden and Finland the ratio is about three to one. Relative to families in the middle of the distribution (at the 50th percentile), low-income persons in the Nordic countries have incomes that are 57% of the median, compared to 47% in Canada and the United Kingdom, and 39% in the United States.
In Table 2, we show how inequality has evolved in these countries since the late 1970s. Perhaps the most remarkable feature is the stability in the relative ranking of countries despite change. Countries with higher levels of family income inequality (including Canada) at the beginning of the period also had higher levels at the end of the period. The Nordic countries (Finland, Sweden) retained their distinctively low levels of inequality through the mid-1990s, at which point they began to move toward Central European levels. Canada’s position relative to the United States and the United Kingdom, however, has changed quite dramatically since the 1970s. At the beginning of the period, the difference between Canada and the United States was quite modest, and Canadian inequality was higher than in the United Kingdom. By the end of the period, however, Canadian inequality was well below US and UK levels, a result of relative stability in the Canadian distribution compared to a long-term rise in inequality in both the United States and the United Kingdom. The relative stability in the Canadian income distribution through the 1980s and early 1990s has been well documented (Beach and Slotsve, 1996; Wolfson and Murphy, 1998; Gottschalk and Smeeding, 1997) and reflects the fact that, through the 1980s and early 1990s, increases in market-based inequality among families were offset by rising social transfers. As we show in the following section, however, new evidence from recent studies indicates that income inequality in Canada deviated from this long-standing stable trend and increased moderately during the 1990s.
Is Family Income Inequality Rising in Canada?
The data in Table 2 show that income inequality, as measured by the Gini coefficient, changed little between the late 1970s and the late 1990s. These results are based on a survey that was terminated in 1997 in favour of a new survey. However, to assess recent trends one should analyze comparable points in the business cycle, and employ more recent data. To do so, we turn to two sources: the census, and a series that is a combination of the terminated Survey of Consumer Finances (SCF) and its replacement, the Survey of Labour and Income Dynamics (SLID). We focus on 1990 and 2000, roughly the last two business cycle peaks.1
The survey data (SCF plus SLID) registered a 6% increase in the Gini coefficient between 1990 and 2000, based on after-tax and transfer income (i.e., disposable income2), the most appropriate income measure. However, the census does not collect taxes paid data, so we turn to a slightly different definition of family income: income after government transfers and pre-tax. On that basis, the survey data again suggest a 6% increase in the Gini between 1990 and 2000, and the census data registered a 5% increase. Thus, these two sources indicate a moderate increase in income inequality over the 1990s.
Perhaps a more intuitive interpretation of the increase can be gained using top to bottom ratios. Based on survey (SCF and SLID) data, after taxes and transfer income among families in the top decile (the 10% of families with the highest incomes) was 7.5 times that of the income of families in the bottom decile in 1990. By 2000, this had increased to an estimated 8.7 times greater,3 an increase of 15% in the top to bottom ratio. Census data suggested a similar 14% increase in this ratio (based on pre-tax, after transfer income).
Figure 1 indicates the rise in inequality was primarily the result of faster rising incomes at the top of the income distribution. Census data suggest little change in family incomes among lower-income families between 1990 and 2000, while higher-income families saw increases between 7% and 16%.
Measures of income inequality indicate changes in relative distribution of income over the entire population. Low-income statistics, in contrast, provide a more focused view of changes at the bottom end of the income distribution. The low-income rate measures the proportion of people below a low-income cut-off, while the low-income gap is a measure of the depth of low income among those who fall below the cut-off.4 Analysts concerned with economic exclusion typically focus on both measures, as we will see when we turn to our review of Canada-specific trends. Because of measurement problems, however, the cross-national comparisons in the section below only consider differences in low-income rates.
FIGURE 1 Percent Change in Total Family Income by Vingtile,a 1990-2000, Census Data
Source: Statistics Canada, Census.
Low Income in Comparative Perspective
a. Levels and Trends
For cross-national comparisons, we follow conventional practice and measure low income as including all persons in families with incomes less than 50% of the national median (Table 3). Our results are based on Smeeding (2003).
The low-income rate for all persons, using the 50% threshold, varies from 5.4% in Finland to 17.0% in the United States. Higher rates are found in predominately English-speaking nations, countries that also have higher levels of overall inequality. While overall rates in Canada and the United Kingdom (12%) remain well below those of the United States, Canada’s relative position varies substantially among sub-groups. At 5%, the low-income rate among Canadian seniors is now among the lowest of the affluent democracies. In contrast, until the late 1990s, low-income rates among Canadian one-parent families (39%), and childless households (12%) were close to or above US levels. As we highlight below, however, these historically high levels of low income among Canadian lone-parent families moved down sharply after 1997.
|Nation (year)||Overall||Families with Children||Childless||Elders|
|1 Parent||2 Parents|
|United States (00)||17.0||41.4||13.1||11.1||28.4|
|United Kingdom (99)||12.3||31.3||8.9||7.7||24.6|
Source: Smeeding (2003).
|Smeeding (2003) Percent Reduction||Corak et al. Percent Reduction|
b. The Role of Transfers and Taxes
The extent to which income transfers (and taxes) redistribute income to the benefit of the least advantaged varies significantly among advanced economies. It is worth noting that countries differ in the extent to which services, such as education or health care, are provided publicly or privately so estimates, such as those presented below (i.e., based on cash benefits alone) do not capture the full range of government redistribution.
The standard approach to measuring the effect of transfers and taxes, as employed here, is to compare the low-income rate on a pre-transfer/tax basis (market income) with the lowincome rate after transfers and taxes (disposable income).5
Table 4 summarizes the overall impact of taxes and transfers (including those for retirees) on low-income levels from Smeeding (2003) and Corak et al. (2003). Despite differences in data sources, the results are remarkably consistent.6 Among the eight countries, the cash tax/transfer system in the United States reduced low income the least. Reductions were somewhat higher in Canada and the United Kingdom than in the United States, but generally less than in the European countries.
Low-Income Rates in Canada
a. Changes in Canadian Low-Income Levels during the 1990s
In this section, the low-income rate is the proportion of people below Statistics Canada’s after-tax and transfer low-income cut-off (LICO). As shown in Figure 2, the low-income rate typically tracks the unemployment rate. Rates tend to rise during recessions and fall again as recovery sets in; however, in the 1994 to 1997 period, the low-income rate in Canada rose despite improving economic conditions. Two papers (Osberg, 2000; Picot et al., 2003) concluded that this deviation was related to the fact that, despite recovery, earnings did not increase to any great extent among poorer families, while social transfers (social assistance, Employment Insurance benefits) fell. By 2001, though, Figure 2 suggests the low-income rate and the unemployment rate had resumed their historical relationship.
Strikingly, after years of relative stability, low-income rates among lone-parent families fell substantially in the latter part of the 1990s. Statistics Canada calculations show that the low-income rate among lone parents fell 10 percentage points (from 42% to 32%) between 1997 and 2000.7 Though still not well understood, this change was at least in part the result of increased earnings. Average market income among female lone-parent families rose 46% between 1996 and 2001, a trend that appears to be primarily the result of higher labour force participation. About 82% of female lone-parent families had earnings in 2001, while in 1993 the corresponding figure was 67%. This increase may reflect changes in child tax benefits (the National Child Benefit), which reduced employment disincentives, and certain “re-investments” of transfers from social assistance to cash and non-cash programs for children. In effect, earnings trends among loneparent families were a significant and important exception to the patterns reported for all low-income families described earlier.
The low-income gap provides additional information on the well-being of low-income families. Sometimes referred to as the depth of low income, it is the difference between the low-income cut-off and the average income of low-income families expressed as a ratio of the cut-off.
The change in the low-income gap is shown in Figure 3 by indexing the gap to 100 in 1980. Despite falling employment earnings among lowincome families during the 1980s, the low-income gap fell as a result of rising transfers (Picot et al., 2003). From 1993 to 1998, in contrast, average real incomes among low-income families fell resulting in a rising low-income gap.8 As a result, unlike the trend in the low-income rate, the low-income gap did not return to the levels observed at the end of the 1980s.
FIGURE 2 Low-Income Ratea and Unemployment Rate (Left Axis), Canada, 1980 to 2002
Source: Statistics Canada, Census.
FIGURE 3 The Low-Income Gap,a 1980 to 2001 Index, 1980=100
FIGURE 4 Low-Income Rates of Immigrants Relative to Non- Immigrants, by Five-Year Period of Immigration, 1980 to 2000
Source: Picot and Hou (2003).
b. Low-Income Trends Among Recent Immigrants
Low-income trends among the population as a whole tend to mask an underlying divide that has opened up between the native-born and immigrants to Canada. The decline in earnings among successive groups of immigrants entering Canada through the 1980s and 1990s has been well documented (Bloom and Gunderson, 1991; Reitz, 2001; Green and Worswick, 2002; Frenette and Morissette, 2003; Aydemir and Skuterud, 2004).
This rising earnings gap between recent immigrants and the native-born is reflected in a growing divide in lowincome rates. For most major groups among the native-born, low-income rates have been falling through the 1980s and 1990s. In contrast, lowincome rates among immigrants (net of business cycle effects) have been rising. Between 1980 and 2000, two years of business-cycle peaks, the lowincome rate among recent immigrants rose from 25% to 36%, while falling among the Canadian-born from 17% to 14%.9 Even among traditionally vulnerable groups of the Canadianborn, such as seniors, lone parents, and young families, low-income rates either fell, or remained constant (Picot and Hou, 2003). Among very recent immigrants (in Canada five years or less), the low-income rate was 1.5 times that of the Canadianborn in 1980, rose to 2.7 times the native-born rate in 1995, and declined marginally to 2.5 times the nativeborn rate in 2000.
With the advent of longitudinal household surveys, analysts are now able to establish levels and trends in the low-income rate and gap, and patterns of entry and exit from low income, the duration of low-income spells, and the movement of individuals within the income distribution. Income dynamics analysis is concerned with the fluidity of movement among various income states and the determinants of these flows.
a. Low-Income Dynamics in a Comparative Context
For many, low income is a transient condition. Corak et al.’s analysis of low-income dynamics in Canada, Germany, the United Kingdom, and the United States shows that between one third and one half of all people who enter low income exit after one year (Table 5).10 At the other extreme, between 20% and 30% of all people who enter low income find themselves still in that state after five years. Though limited in its comparative scope to only four countries, the results of Corak et al.’s study indicate that national differences in persistent low income are correlated with higher inequality and low-income levels. Of those entering low income in the United States, 31% were still in the state after five years, compared to 24% in Canada, and 18% in Germany and the United Kingdom. Since the number of people persistently in low income depends on how long they remain in that state and how many enter, a somewhat better indicator of long-term exclusion is the percentage of the total population in low income over an extended period. During the six years analyzed by Corak et al., 5.4% of the population was in low income in all six years in the United States, 4.4% in the United Kingdom, 2.9% in Canada, and only 1.9% in Germany.
b. The Most Vulnerable: The Concentration of Persistent Low Income in Canada
If persistent low income is highly concentrated among a few groups, then research focusing on the outcomes of these groups will be useful. Entering and exiting low income is not a simple matter of finding or losing jobs. Family formation patterns (divorce, separation, marriage, common- law unions) often play a major role. And regarding employment outcomes, the underlying barriers to improved outcomes may vary tremendously among groups. These barriers and underlying causes of poor employment outcomes are almost certainly very different for, say, recent immigrants, than for single parents or persons with work disabilities.
Based on SLID, Michael Hatfield and his colleagues at Human Resources Development Canada attempted to identify those groups that account for a disproportionate share of persistent low income (Hatfield, 2003). They found that while accounting for only 26% of the population, 62% of all persons in persistent low income over the 1996-2000 period were in one of five groups (Table 6): lone-parents, unattached persons aged 45 to 64, recent immigrants (in Canada 10 years or less), persons with work-limiting disabilities, and Aboriginal people living off reserves.11 Persons with work-limiting disabilities constituted the single largest group (26%) of those persistently in low income over the period, followed by recent immigrants and unattached individuals aged 45 to 64.
|United States |
|A. Leaving low income after one year|
|B. In low income after five years|
|C. Of the population in low income at least once|
|D. Of the population in low income in all six years|
a. 1993 to 1998 (1990 to 1996 for the United States).
Source: Corak et al. (2003), based on the Survey of Labour and Income Dynamics for Canada, the British Household Panel Survey, the German Socio-Economic Panel, and the Panel Survey of Income Dynamics in the United States.
|Percent of Population 16-64|
|Low-Income in 2000 |
|Low Income Measured Over Five-Year Basis (persistent) 1996-2000 |
|Members of at least one of the five groups||25.9||56.2||62.1|
|Recent immigrants (10 yrs)||5.5||11.7||16.6|
|Work – limiting disability||10.7||22.6||26.3|
|Off-reserve Aboriginal people||3.1||5.1||5.5|
|Others not in the five groups||74.1||43.8||37.9|
Source: Hatfield (2003), based on Survey of Labour and Income Dynamics.
|% of Group in Low Income|
|2000||Persistent over 1996-2000|
|Recent immigrants (10 yrs)||22.8||24.2|
|Self-identified Aboriginal people (off-reserve)||17.4||17.2|
Source: Hatfield (2003), based on Survey of Labour and Income Dynamics.
It would be incorrect to think that all persons in these five groups find themselves at the bottom of the income distribution for extended periods. Even among these groups, a minority is exposed to low income on a long-term (or even transient) basis (Table 7). Roughly a quarter to a third of the members of these groups found themselves in persistent low income over a five-year period.12 Though not in the majority, persistent low income in these populations is about eight times the average of 4.2% found in the rest of the population.
There is a substantial debate regarding the extent to which the concentration of persistent low income should translate into a strategy of targeting policies. Focusing policies on specific groups could, by definition, exclude many others who may be in exactly the same economic position, but who would not be eligible for program assistance. Furthermore, broad support for redistributive programs may become problematic when eligibility is determined in part by group membership rather than by some measure of need. Using the information on group concentration to determine underlying causes of persistent low income, and developing policy strategies that are then made available to the entire population may be the most fruitful approach.
Summary and Conclusion
The standard conclusion of the research literature on Canadian incomes through the mid-1990s was that, despite some worrying developments, Canada had largely avoided the sharp rise in income inequality evident in both the United States and the United Kingdom since the mid- 1970s. The worrying developments included the declining earnings of younger adults (under 35), particularly males, the corresponding impact of this development on young families, and falling earnings among recent immigrants and less educated males. Changes in earnings combined with changing patterns of labour market participation and family structure produced a rise in inequality in market incomes (earnings) among families. However, through to the mid-1990s, most if not all of this change was offset by rising income transfers.
Recent evidence indicates this trend began changing in the 1990s. The gains associated with the economic expansion of the ’90s went mainly to higher-income families while the earnings of poorer families stagnated and social transfers fell. The result was a moderate increase in family income inequality.
At any point in time, persistent low income is concentrated among five groups. When focusing on increases in low-income rates, the focus shifts squarely to recent immigrants. While low-income rates have been falling among most Canadian-born groups, rates have been rising rapidly among recent immigrants. This deterioration in family welfare is related to the declining earnings of recent immigrants during the 1980s and 1990s despite their very high educational qualifications.
In contrast to these bad news stories, there are a few good news stories to report. The dramatic decline in lowincome levels among Canadian seniors since the 1970s, a result of the maturation of legislative changes introduced in the 1960s, has been evident and well documented for some time. The recent and substantial decline in low-income rates among lone-parent families is less well understood. The decline appears to be the result of higher rates of labour force participation among lone parents, possibly reflecting changes in the family benefit system aimed at breaking down the work disincentives built into traditional social assistance schemes. Why this shift occurred and its future evolution are topics deserving considerable attention. Other good news stories relate to the rising earnings of women (particularly the highly educated), older workers, and higher-income (and education level) families.
In comparative terms, the mid-Atlantic metaphor – somewhere between the United States and Europe – has often been used to describe Canadian levels of income inequality, low income, and social spending. In the 1970s, the metaphor was somewhat misleading since Canadian levels of low income and inequality left Canada anchored in close proximity to the American shoreline. Indeed prior to the mid- 1970s, low-income rates among children and seniors were higher in Canada than in the United States. Since then, the metaphor has become somewhat more accurate, not so much because Canada has moved closer to Europe, but because the United States, now joined by the United Kingdom, has moved further away from Europe.
Aydemir, A. and M. Skuterud. 2004. “Explaining the Deteriorating Entry Earnings of Canada’s Immigrant Cohorts.” Analytical Studies Branch Research Paper #225, Analytical Studies, Ottawa: Statistics Canada.
Beach, C., and G.A. Slotsve. 1996. “Are We Becoming Two Societies? Income Polarization and the Myth of the Declining Middle Class in Canada.” The Social Policy Challenge 12. Toronto: C.D. Howe Institute.
Bloom, D.E., and M. Gunderson. 1991. “An Analysis of the Earnings of Canadian Immigrants.” In Immigration, Trade and the Labour Market, eds. John M. Abowd and Richard B. Freeman. Chicago: The University of Chicago Press.
Corak, M., W-H. Chen., A. Demanti, and D. Butler. 2003. “Social Cohesion and the Dynamics of Income in Four Countries.” Analytical Studies Branch Research Paper Series, Analytical Studies, Ottawa: Statistics Canada.
Frenette, M., and R. Morissette. 2003. “Will They Ever Change? Earnings of Immigrant and Canadian Born Workers Over the Last Two Decades.” Analytical Studies Branch Research Paper #215, Analytical Studies, Ottawa: Statistics Canada.
Gottschalk, P., and T.M. Smeeding. 1997. “Cross-National Comparisons of Earnings and Income Inequality.” Journal of Economic Literature XXXV (June): 633-687.
Green, D.A., and C. Worswick. 2002. Earnings of Immigrant Men in Canada: The Roles of Labour Market Entry Effects and Returns to Foreign Experience. University of British Columbia, Department of Economics, prepared for Citizenship and Immigration Canada.
Hatfield, M. 2003. “Persistent Low-Income: A Key Barrier to Social Inclusions.” Applied Research Branch, Human Resources Development Canada. Mimeo.
Osberg, L. 2000. “Poverty in Canada and the USA: Measurement, Trends and Implications.” Presidential Address to the Canadian Economics Association, Vancouver, June 3, 2000.
Picot, G., and F. Hou. 2003. “The Rise in Low-Income among Immigrants in Canada.” Analytical Studies Branch Research Paper Series 11F0019MIE2003198, Analytical Studies, Ottawa: Statistics Canada.
Picot, G., R. Morissette, and J. Myles. 2003. “Low-Income Intensity during the 1990s: The Role of Economic Growth, Employment Earnings and Social Transfers.” Analytical Studies Branch Research Paper Series 11F0019MIE2003172, Analytical Studies, Ottawa: Statistics Canada.
Reitz, J. 2001. “Immigrant Success in the Knowledge Economy: Institutional Changes and the Immigrant Experience in Canada, 1970-1995.” Journal of Social Issues 57: 579-613.
Smeeding, T. 2003. “Government Programs and Social Outcomes: The United States in Comparative Perspective.” Mimeo. Centre for Policy Research, Maxwell School, Syracuse University.
Wolfson, M., and B. Murphy. 1998. “New Views on Inequality Trends in Canada and the United States.” Analytical Studies Branch Research Paper Series, No. 124, Analytical Studies, Ottawa: Statistics Canada.
The actual top to bottom ratio in SLID in 2000 was 9.3. However, the move from the SCF data to SLID data in 1996 resulted in the top to bottom ratio artificially increasing by 0.6. Hence, to make the 2000 ratio comparable to the 1990 ratio, we estimated that the 2000 ratio would have been 8.7 (9.3-0.6) had there been no change in data sources.
Such comparisons only measure the direct effects of income transfers however, and possible behavioural (i.e., indirect) responses are not considered. For example, reductions in transfer benefits may be an incentive for people to seek employment, thus raising earnings levels.
Corak et al. (2003) drew on longitudinal data to study income dynamics, and the Smeeding paper (2003) relied on crosssectional data from the Luxembourg Income Study, an organization dedicated to obtaining comparable income data for numerous western industrialized countries.
To maintain consistency with the lowincome concept used in the comparative tables, figures are reported based on the LIM low-income cut-off (i.e., persons in families whose adjusted income is less than 50% of median income). Results based on the more usual LICO-based measures, however, tell an identical story.
The low-income cut-off used in this analysis is the Statistics Canada aftertax/ transfer LICO, but it is held constant (except for adjustments to account for changes in inflation) over this period. Hence, an increasing gap means average real family incomes were falling among low-income families.
The low-income rate among recent immigrants fell between 1995 and 2000 from 47.0% to 35.8%, but this was primarily associated with the significant improvement in economic conditions that caused the low-income rate for all groups to decline. Abstracting from business cycle effects (rates rise in recessions and fall in recoveries), the low-income rate among recent immigrants has been increasing.
Bookmark - The Face of Low Income In Canada’s Metropolitan Areas
Across the nation, businesses, policy makers, and Canadians from all walks of life share a heightened interest in, and awareness of, the state of Canada’s metropolitan areas. They are concerned about renewing community life in urban centres. This means, among other things, addressing poverty, providing new opportunities to learn and work for all Canadians, including new immigrants and Aboriginal people, and enhancing the business climate.
Recently, Statistics Canada released the first in a series of reports examining trends and conditions in Canada’s largest urban areas. Using census data, Low Income in Census Metropolitan Areas examines income and low income in Canada’s 27 urban centres between 1980 and 2000. The report observes that, while the low-income rate did not rise in most of Canada's large urban centres between 1980 and 2000, there was a dramatic change in the demographic make-up associated with a low income.
For more information on this topic, please see the Statistics Canada report, Low Income in Census Metropolitan Areas: 1980 to 2000, by Andrew Heisz and Logan McLeod. The full report is available on-line at <www.statcan.ca> catalogue number is 89-613-MIE, No. 001. Other reports in this series, covering demographics, immigration, housing, health, labour markets and industrial structure, work location and commuting mode, Aboriginal people, and culture, are now, or soon will be, available.