Introduction
The rapid decline in poverty among older people in Canada since the 1970s often is linked to the development of state social policies (Myles, Reference Myles2000; Smeeding & Sandström, Reference Smeeding and Sandström2005; Turcotte & Schellenberg, Reference Turcotte and Schellenberg2007). Yet poverty remains a critical social problem, especially among select subpopulations (Gazso, Reference Gazso2005; Veall, Reference Veall2008). Elderly immigrants, particularly the recently arrived, have been identified as one of the most vulnerable of these subgroups (Alternative Planning Group, 2009; Baker, Benjamin, & Fan, Reference Baker, Benjamin and Fan2009). In 2006, about 30 per cent of immigrants aged 65 and older who were in Canada fewer than 20 years were living in households having incomes below Statistics Canada’s Low Income Cut-off.Footnote 1 By contrast, the low-income rates of the overall elderly population in Canada on average dropped 20 percentage points from 1980 to 2003, becoming stable at around 15 per cent in the early 2000s (Turcotte & Schellenberg, Reference Turcotte and Schellenberg2007).
The different outcomes by nativity suggest that state social policies may unequally support older people, making immigrant seniors more vulnerable to social risks such as poverty. At the same time, responsibilities for the sustenance of the economic well-being of older people also are shared by the family (Jenson, Reference Jenson2004). This, too, has implications for elderly immigrants, particularly for recent arrivals who may have limited access to government income support systems. To these seniors, family assistance can play a significant role in alleviating poverty.
Although the roles of state social policies and family assistance are of long-standing interest in the study of elderly poverty, the joint impacts of the state and family on the poverty levels of elderly immigrants have received less attention. To remedy this research gap, we conducted a study of the poverty of elderly immigrants, focusing on the combined impacts of state support and family assistance.Footnote 2 More specifically, we asked: to what extent does the poverty of elderly immigrants reflect state policies, and to what extent is this lack of state intervention ameliorated by family financial assistance? In answering this question, we posed three sub-questions. First, how does the poverty of elderly immigrants compare with that of the Canadian-born elders? Second, is the (expected) income disadvantage of immigrant elders relative to their Canadian-born counterparts explained by a lack of government support, and if so, to what extent? Third, to what extent is the poverty of elderly immigrants alleviated by financial help from their families? We used data from the 2006 Canadian Census of Population to address these questions.
In the analysis, we included gender and ethnicity of immigrant and Canadian-born elders to draw a nuanced picture of elderly poverty. It is widely known that older women are at higher risks of living in poverty than are their male counterparts, especially in later stages (e.g., age 75 and older) of their lives (Smeeding & Sandström, Reference Smeeding and Sandström2005). Their vulnerability to poverty is often attributed to their propensity to live alone. Our research considered whether this argument especially applies to elderly immigrant women who are known to receive much family financial support by co-residing with their children.
This study also took into accountthe growing ethnic and racial diversity among older people (both Canadian-born and foreign-born). Such diversity reflects temporal changes in waves of immigration throughout the 20th and 21st century, first from European countries and then from other world regions. Such changes began in North America with legislative changes in the 1960s that removed national origins as the main criterion of admissibility, replacing it with the criteria of family reunification, humanitarian concerns, and economic contributions. By the start of the millennium, well over half of immigrants in Canada and in the United States were from countries and regions other than Europe, the latter including the United Kingdom and the Republic of Ireland. The ethno-racial composition of elderly immigrants also bears the imprint of such changes, both because immigrants arriving as adults “age in place” and because they immigrate in old age to rejoin family members (Boyd, Reference Boyd1991).
In addition to the impact of demographic changes in source countries, two additional rationales prompted our study of ethno-racial variations in the poverty of elderly immigrants. First, the tendency to receive family assistance varies among ethno-racial groups, reflecting cultural norms regarding elder care (Kaida, Moyser, & Park, Reference Kaida, Moyser and Park2009). Second, our study contributes to the ongoing discourse over the persistence of ethnic and racial stratification into old age. Historically, the two founding charter groups, the British and the French, dominated both politically and in terms of economic status, creating an ethnically stratified society. However, some researchers have now argued that the traditional hierarchy of ethnic differences has been replaced by race and color differences (Lian & Matthews, Reference Lian and Matthews1998). Whether race and color trump European origins among Canada’s elderly population is an empirical question addressed in our research. We suspected that it might, if only because new immigrant groups increasingly are persons of color, and as recent arrivals, they are the most exposed to state regulations that prorate income assistance. Moreover, racial inequalities in the labor market (e.g., occupational segregation, pay gaps) may lead to racial income inequality in old age by reducing access to retirement incomes, including the Canada/Quebec Pension Plans and occupational pensions.
To this end, we compared the poverty levels of over 30 ethnic and racial-immigrant status groups, taking advantage of the detailed information on ethnic origins available in the 2006 Census data. Our study is an improvement on existing U.S. studies of ethno-racial variations in elderly poverty that tend to use more aggregated ethno-racial categories, such as “Asians” and “Whites” (Elmelech & Lu, Reference Elmelech and Lu2004; Rendall & Speare, Reference Rendall and Speare1995; Willson & Hardy, Reference Willson and Hardy2002).
State Support for Elders
In welfare states such as Canada, governments play key roles in income security for older people, particularly when they retire from the paid labor force (Lewin & Stier, Reference Lewin and Stier2003). Government social policies for the income security of older people typically are of three kinds: (a) universal pensions, (b) social insurance, and (c) social assistance (Smeeding & Sandström, Reference Smeeding and Sandström2005). For the most part, universal pensions are a “pay-as-you-go, flat-rate benefit” (Smeeding & Sandström, Reference Smeeding and Sandström2005), while social insurance is closely tied to the earnings histories of older individuals. Social assistance targets elders in the bottom rungs of income distributions through income supplements, usually provided by sub-national units of governments.
These three income maintenance policies are found in Canada’s pension plans for seniors. First, universal pensions are available in the form of Old Age Security (OAS) – a non-contributory pension provided to any Canadian citizen or legal resident aged 65 and older who has lived in Canada for 10 years or more after age 18 (Dempsey, Reference Dempsey2005; Service Canada, 2011). As a supplement of the OAS, the Guaranteed Income Supplement (GIS) may be provided to OAS recipients with limited or no other income. The second type of income maintenance policy is social insurance, which consists of two major pension sources: (a) the Canada/Quebec Pension Plans (C/QPPs); and (b) occupational pensions and the Registered Retirement Savings Plans (RRSP). The C/QPPs are “supplied to past members of the labour force [who have] contributed to the plans alongside their employers” (Gazso, Reference Gazso2005). In order to qualify for the full C/QPP benefits, one must contribute for 40 years. Occupational pensions and RRSPs can be “sponsored by employers, labour unions, and professional organizations” by deferring wages to provide for the employees’ retirement (Gazso, Reference Gazso2005). Although occupational pensions and RRSPs are in the private domain, the government’s tacit involvement remains, given that contributions are deducted from taxable income. The third type of income maintenance policy is social assistance, for which economically constrained seniors can be eligible as part of supplementary income provided by provincial governments (Dempsey, Reference Dempsey2005).
Canada’s state support system for older people is considered successful in reducing elderly poverty (Turcotte & Schellenberg, Reference Turcotte and Schellenberg2007). In particular, the Canadian government’s prioritization of universal pensions is credited with reducing elderly poverty and with making the application for OAS/GIS simple and less stigmatized (Smeeding & Sandström, Reference Smeeding and Sandström2005). However, despite the overall success in reducing elderly poverty, research shows that disparities persist in the receipt of state support among seniors in Canada. Notably, immigrants, women, and ethnic minorities often are subject to institutional barriers which constrain their receipt of state support (Angel, Reference Angel2003; Boyd, Reference Boyd1991). The following brief discussion shows how existing state support systems influence the receipt of support by immigrant elders (as well as elderly women and minority elders) in the three government strategies just described (public pensions, social insurance, and social assistance).
One way in which state support systems influence support is that the 10-year residency requirement for OAS may deter immigrant seniors from benefiting from Canada’s universal pension plans; this requirement particularly affects recently arrived immigrant seniors who have not accumulated years of residence to be eligible for OAS (Boyd, Reference Boyd1991; Marier & Skinner, Reference Marier and Skinner2008). Second, Canada’s social-insurance-premised programs, such as C/QPPs and occupational pensions, are problematic for elderly immigrants because of contribution requirements and disparities in availability for people in the labor force. With their relatively shorter period of participation in the Canadian labor market, elderly immigrants are less likely to receive C/QPPs because they are less likely to meet the pensions’ long-term contribution requirement. In addition, as the literature on precarious employment suggests, immigrants are less likely to find jobs that are full time, permanent, in the public sector, and unionized; this implies that immigrants are less likely to receive social wages from employment, including private occupational pensions (Cranford & Vosko, Reference Cranford, Vosko and Vosko2006; Kalleberg, Reference Kalleberg2009). Moreover, inequalities faced by women and minorities throughout the life course make these vulnerable groups even more economically vulnerable in their old age (Angel, Reference Angel2003; Cheal & Kampen, Reference Cheal and Kampen1998; Gazso, Reference Gazso2005; Marier & Skinner, Reference Marier and Skinner2008; Rupp, Strand, & Davies, Reference Rupp, Strand and Davies2003; Wanner & McDonald, Reference Wanner and McDonald1986). Women and minorities are more likely to encounter occupational segregation and pay gaps, and women are often burdened with unpaid work responsibilities at home, suggesting a weaker labor force attachment and less access to work-related pensions (e.g., the C/QPPs, occupational pensions) after labor force disengagement.
A third way in which state support systems influence support is that elderly immigrants, particularly recent arrivals, are affected by the government’s regulations governing access to social assistance. As part of the “undertaking” contract with the federal department mandated to admit immigrants, Citizenship and Immigration Canada, sponsored immigrants, many of whom are elderly, are not eligible for social assistance, including income security programs, for up to 10 years after arrival (Baker, Benjamin, & Fan, Reference Baker, Benjamin and Fan2009; Boyd, Reference Boyd1991; Marier & Skinner, Reference Marier and Skinner2008). Their sponsoring families are obliged to provide financial support during this period. In addition, immigrant seniors tend to encounter difficulties in accessing government support – difficulties that may be related to their migration experience. Language barriers and lack of information are deemed major obstacles in accessing social support even when an elderly immigrant is eligible for the support (Dietz, Reference Dietz2001).
Family Support for Elders
As discussed previously, the state is often considered a key player in the overall reduction of elderly poverty in industrialized countries including Canada. Yet there is reason to suspect that the family is the primary institution in the private sphere serving the economic well-being of older people (Burr & Mutchler, Reference Burr and Mutchler1999). Elderly immigrants, as well as other subgroups of elders, including women and ethnic minorities, remain vulnerable to poverty partly because of their limited access to state support. For these subgroups, family financial assistance can play a major role in alleviating poverty.
In fact, family financial support has significant implications on the economic well-being of elderly immigrants in Canada. As mentioned earlier, the Canadian government mandates families’ financial support when they sponsor elderly relatives from abroad by prohibiting the receipt of government income assistance during elderly persons’ first 10 years of residence in Canada. Thus, for elders who have recently immigrated to Canada through family reunification, children and other relatives (e.g., grandchildren, siblings) are the primary source of income, unless they have income from employment, and this is difficult for newcomers in the host country’s labor market.
While family support is deemed crucial for the alleviation of elderly poverty in theory, empirically proving its impact is no easy task.Footnote 3 Information on income received from an elderly person’s family is not readily available in most survey data. Given such data limitations, researchers often use elderly individuals’ living arrangements as a crude measure of the income received from kin (Rendall & Speare, Reference Rendall and Speare1995), even though this may lead to measurement errors (for details, see the Methods section). More specifically, for elderly persons, living with kin and forming a large household means the pooling of income as a household, and co-residence may be an effective means of escaping poverty by taking advantage of the economy of scale (Chesley & Poppie, Reference Chesley and Poppie2009; Picot, Lu, & Hou, Reference Picot, Lu and Hou2009). In fact, Rendall and Speare (Reference Rendall and Speare1995) found that co-residence with extended families reduced elderly poverty in the United States by as much as 42 per cent. There are no comparable studies in Canada. However, in their review of low-income trends among immigrant seniors, Picot et al. (Reference Picot, Lu and Hou2009) conjectured that the temporal declines over the past 25 years could partly reflect the observed temporal increases in the tendency of the foreign-born elderly to live in multigenerational households.
Our present study adopted Rendall and Speare’s (Reference Rendall and Speare1995) approach and used living arrangements as a measure of family support. We expected that poverty among immigrant elders was alleviated to some degree because of their patterns of living arrangements (i.e., higher propensities towards co-residence with kin), even though their actual poverty levels were high.
In this article, we also considered ethnic-racial and gender variations in the propensity towards co-residence with kin among the elderly. There is ample evidence in the literature on aging that co-residence with kin is prevalent among some ethno-racial groups in North America, including Hispanics, Blacks, Chinese, Greeks, Italians, Portuguese, and South Asians, particularly among women (See Angel & Tienda, Reference Angel and Tienda1982; Burr & Mutchler, Reference Burr and Mutchler1993, Reference Burr and Mutchler1999; Lee & Angel, Reference Lee and Angel2002; Wilmoth, DeJong, & Himes, Reference Wilmoth, DeJong and Himes1997, for U.S. cases; see Gerber, Reference Gerber1983; Kaida, Moyser, & Park, Reference Kaida, Moyser and Park2009; Ng, Northcott, & Abu-Laban, Reference Ng, Northcott and Abu-Laban2007; Thomas & Wister, Reference Thomas and Wister1984, for Canadian cases). In contrast, independent living (e.g., living only with spouse, living alone) is more prevalent among Northern Western European groups, including British, German, and Dutch origin groups.
Quantitative research in aging has sought to rigorously explain such ethno-racial variations in elderly living arrangements over the past two decades. Two streams of explanations dominate the literature. The first explanation is that the different patterns of living arrangements among elders reflect cultural preferences; elders from cultures that emphasize family-oriented values are more likely to live with their children or other relatives, whereas elders from cultures that emphasize an individual’s privacy tend to live independently. The second explanation posits that co-residence with kin is a strategy to cope with economic constraints – a strategy that prevalently is adopted by older people who lack economic resources due to their disadvantageous standings in the society (e.g., ethno-racial minorities, women). A recent study by Kaida et al. (Reference Kaida, Moyser and Park2009) has found that both cultural preferences and economic constraints influence variations in living arrangements among the elderly of the seven select ethnic groups in Canada (British, Dutch, Germans, Italians, Chinese, East Indians, and South Asians). However, the relative importance of these two explanatory factors varies by marital status. Cultural preferences explain variations in living arrangements among the elders from these seven ethnic groups to a greater extent for married elders, whereas economic constraints explain those variations among non-married elders.
We considered such known variations by splitting the elderly population (both immigrants and the Canadian-born) by ethno-racial origin. We expected that some groups (e.g., Italians, Portuguese, Greeks, Chinese, South Asians) known for their higher propensity towards co-residence with kin were perhaps protected from poverty to some degree. Put differently, if elders of these ethnic groups had the same distributional characteristics of living arrangements as the mainstream elderly population (Canadian-born of British descent), their chances of falling into poverty would be higher. We performed analysis separately by gender.
Methods
Data and Sample
To address our research questions, we analyzed data from the 2006 Census Public Use Micro-data File (PUMF) of individuals.Footnote 4 In this database, a 2.7 per cent sample of the population was enumerated in the 2006 census of population with information on approximately 844,000 individual respondents of all ages. Unlike the U.S. census public use file, it was not amenable to the construction of a hierarchal file in which all members of a family unit could be identified. Although two separate family and household files also were produced by Statistics Canada for earlier censuses, the data on these files were highly aggregated, preventing their use in our study.
We restricted our sample to non-Aboriginal persons aged 70 and older, who were permanent residents of Canada, and who were members of private households. We selected persons aged 70 and older at the time of the census (May 2006) for two reasons. First, we needed to ensure that everyone in our sample had fully turned 65 and older by January 2005 and was eligible to receive the government retirement incomes (e.g., Old Age Security) throughout 2005 (total family income for 2005 was used by the census to determine low-income status). This requirement automatically restricted the age group to those aged 67 and older at the census date. Second, the 2006 Census PUMF coded age in five-year intervals. This practice meant that we could not select those aged 67 and older; instead, our analysis focused on those aged 70 and older.
Permanent residents included non-immigrants (Canadian citizens by birth) and immigrants (persons granted the right to live in Canada permanently by immigration authorities). Excluded from the analyses were those who were not permanent residents of Canada (persons from another country admitted temporarily for purposes of employment, study, or who were seeking admission on humanitarian grounds as refugee claimants) and those who resided in collective dwellings.Footnote 5 We also excluded those who resided in the Territories, and those who immigrated in 2005 or 2006. Residents of the Territories were excluded because of the non-existence of information on low-income status for these geographical areas in the Public Use Micro-data File (Statistics Canada, no date). For the same reason, immigrants who arrived in 2005 and 2006 were omitted from our analysis.
In our multivariate analysis, we compared the poverty levels of elders, who were divided into a total of 31 groups by nativity, immigration history, and ethnic-racial origins (for both women and men). To highlight the situations of immigrants versus those of the native-born population, we used responses to the census question on ethnic origins as our measure of diversity rather than birthplace; ethnic data were collected by the census for all respondents unlike birthplace data, which was available only for the foreign born. The Canadian census question probed ancestral origins; however, both historically (Ryder, Reference Ryder1955) and today, religious and racialized labels, such as Jewish, Chinese, Vietnamese, and Black, also were tabulated as valid responses. The comparison of these 31 ethno-racial groups allowed us to consider varied degrees of state support and family support which were perhaps associated with immigration experience and ethno-cultural heritage, as we discussed. Ethno-racial and nativity groups that had 65 (unweighted) observations or more for both women and men stood as separate groups, while groups not meeting this sample size requirement were aggregated as “others” (Table 1, columns 5–6).
Table 1: Percentage and frequency distributions of ethnic-nativity groups, for ages 70 and older
a Percentages may not add up to 100 due to rounding errors.
b Includes Canadian, provincial, or regional origins; Latin, Central, and South American origins; other Western European origins; Swedish, Norwegian, Danish, Finnish, other Northern European origins; Russian, Romanian, other Eastern European origins; Spanish, other Southern European origins; African origins; Lebanese, other Arab origins; Korean, Vietnamese, other East and Southeast Asian origins; all other single-response origins; and multiple origins.
c Includes Canadian, provincial, or regional origins; Latin, Central, and South American origins; other Western European origins; Swedish, Danish, Finnish, other Northern European origins; Hungarian, Russian, Romanian, other Eastern European origins; Greek, Portuguese, Spanish, other Southern European origins; African origins; Lebanese, other Arab origins; Korean,Vietnamese, other East and Southeast Asian origins; and all other single-response origins.
Source: Authors’ calculation from the 2006 Census Public Use Microdata File (PUMF) of individuals
For the foreign-born, after dividing them by ethnic origins, we categorized the ethnic origin groups into two broad groups based on the groups’ migration histories at the time of the 2006 Census: “old” waves and “new” waves of immigration. We used the term wave to refer to a movement of people typically from a specific set of countries in a given period of time. In this article, the old wave refers to the groups that immigrated mostly before the 1970s – groups of immigrants from European countries. In contrast, the new wave referred to those groups immigrating in large scale since the 1970s, following the elimination of the restriction on immigration based on national origin in the late 1960s (Boyd, Reference Boyd1976; Green & Green, Reference Green and Green1999; Marier & Skinner, Reference Marier and Skinner2008). The new waves had predominantly non-European origins, primarily from Asia, but they also included the Caribbean, the Middle East, and Africa. Unlike the United States, Canada has not had a long and sustained history of migration from Mexico or from other Central and South American countries; consequently, the small, current Hispanic populations were collapsed into other single origins or multiple origin groups. Those foreign-born single origin ethno-racial groups (both new and old waves) that had fewer than 65 cases for both women and men were collapsed into “the other single origin group.” The foreign-born with multiple ethnic origins (e.g., Canadian, British, French, and provincial origins only) were also collapsed into the “multiple origin groups.”
Dependent Variable
The focal dependent variable was the respondent’s low-income status. This variable measures the economic well-being of the household in which the individual resided.Footnote 6 The variable was coded as 1 if the total income (before tax) of a respondent’s economic family (or income of a person not in an economic family) in 2005 was below the low-income cut-off point, and otherwise coded as 0 (reference group).
Statistics Canada determines the low-income cut-offs (LICOs) using family size and the size of area of residence. Family size can range from one to seven or more members. The size of area of residence is categorized into five groups: (a) rural (farm and non-farm); (b) small urban regions; (c) a population of 30,000 to 99,999; (d) a population of 100,000 to 499,999; and (e) a population of 500,000 or more. The combination of these two factors produces 35 possible income cut-offs. If the total income of an economic family (or a person not in an economic family) is below the LICO that applies to the family, all family members are considered below the LICO.
Conceptually, LICOs are constructed according to the proportion of annual family income spent on food, shelter, and clothing. An income threshold is set at 20 percentage points higher than the proportion of the income spent on essentials by the average family. Currently, LICOs are based on the 1992 Family Expenditures Survey, which showed that the average family spent 43 per cent of its after-tax income on the aforementioned daily necessities. Therefore, if a family spends more than 63 per cent (43% + 20%) of its after-tax income on food, shelter, and clothing, all family members are considered to be in “straitened circumstances” (Statistics Canada, 2004). Statistics Canada has cautioned that LICOs are not “a de facto definition of poverty”, although researchers often use LICOs in their discussions of poverty in Canada (Statistics Canada, 2004).
Independent Variables
As we have indicated, our study compared the impacts of state and family support for the variations in poverty levels among the elderly. Given the information available in the census PUMF, we used four sources of retirement incomes as proxies of state support. The four sources were (a) OAS/GIS; C/QPPs; (b) other government sources (including all transfer payments; for details refer to Statistics Canada, 2004: 161); (c) retirement pension plans and superannuation, and (d) annuities. We expected that a lack of income from OAS/GIS, C/QPPs, or retirement pensions increased the chance of living in low-income households. By contrast, other government sources of income, such as social assistance payments, were offered on the basis of financial need, and the absence of income from other government sources thus implied economic advantage.
As mentioned earlier, family support was represented by a respondent’s living arrangements, on the assumption that elders living with kin received income support from their co-residing family members, whereas elders not living with kin presumably had no such support. Admittedly, the use of living arrangements as a measure of family support was a risk for making errors in our measurements, since elderly persons perhaps received financial support from their kin, even if they were not in the same household. However, given the absence of information on kin-based income assistance in the census, we argued that living arrangements were the best available measure of family support from census data. We constructed the living arrangement variable from household type, census family type, and marital status variables in the census data. This produced seven mutually exclusive categories: (a) living in a family of a married or common-law couple without children (the reference group); (b) living in a family of a married or common-law couple with children; (c) lone-parent family; (d) multiple-family household; (e) living in a non-family household with relatives; (f) living in a non-family household with non-relatives only; and (g) living in a non-family household alone.
Control Variables
Our multivariate analyses controlled for distributional differences among groups in age, language use, highest level of education, city of residence, and year last worked.Footnote 7 For the elderly population, advancing age could have increased the propensity for low incomes because of dwindling or non-existent retirement funds and/or pensions (Dodge, Reference Dodge1995). Language also was expected to influence elder poverty because of its impact on past labor market experiences. (The ability to use one or both of Canada’s charter languages, English and French, is a form of human capital, enlarging employment opportunities and reducing the likelihood of low income later in life.) We constructed a language use variable from three 2006 Census PUMF variables: (a) the knowledge of official language(s); (b) language spoken most often at home; and (c) language spoken regularly at home. The language use variable consisted of four levels: Level 1 (highest) – knew official language(s), spoke English and/or French most often and regularly at home; Level 2 – knew official language(s), spoke English and/or French most often and spoke other languages regularly at home; Level 3 – knew official language(s), spoke other languages most often, and spoke English and/or French regularly at home; and Level 4 (lowest) – knew official language(s) and spoke other languages most often and regularly at home, or knew neither official language.
Following past studies of elderly poverty, we also controlled for education (Lewin & Stier, Reference Lewin and Stier2003; Rupp et al., Reference Rupp, Strand and Davies2003). It was expected that a lower level of education would result in limited opportunities in one’s life course, which could lead to higher levels of poverty in old age. The impact of local labor markets was captured by city of residence. (Large cities, represented in our study by census metropolitan areas (CMAs), have more extensive knowledge-based economies than smaller towns, and could have offered better employment opportunities to their residents.) Finally, the “year last worked” variable captured the effects of the incidence of unemployment or being out of the labor force.
Analytical Technique
Given that the dependent variable was dichotomous (the total income of an economic family or income of a person not in an economic family being below LICOs versus above LICOs), we employed binary logistic regression to estimate models. We performed logistic regression analyses using a weighting variable that was included on the PUMF to provide population estimates. However, we adjusted this weight by dividing it by a number that represented the unweighted number of cases in the PUMF for the elderly population of interest; this procedure produced normalized weights with a mean of 1. Using normalized weights provided statistical estimates that were representative of the Canadian population of interest but where significance levels were those that would be expected on the basis of the sample size. This avoided overestimating levels of significance for estimates produced in the multivariate analysis. The analyses and the adjustment of the weight variable were conducted separately by gender. Women were more likely to be poor than men, and U.S. research found gender poverty gaps among particular ethnic groups to be related to ethnically based disadvantages faced by females (Elmelech & Lu, Reference Elmelech and Lu2004).
Results
A First Look
Our study focused on a sample of respondents aged 70 years and older. The majority of seniors in our sample were born in Canada, with the Canadian-born of British origins comprising more than 10 per cent of the sample (Table 1, columns 3 and 4). Within the foreign-born, some ethnic minority groups (Germans, Italians, and Chinese) comparatively had large representations, yet none exceeded 10 per cent of the total sample.
Moreover, we found sizeable distributional differences in select characteristics of the Canadian-born, the “old”, and the new waves of immigrants (Table 2, columns 1 to 6). Overall, elders from the new waves of immigration were more likely to lack OAS/GIS, C/QPPs, and retirement pensions than the Canadian-born and the foreign-born from old waves of immigration. In fact, immigrant seniors from old waves showed patterns of the receipt of state support similar to those of Canadian-born seniors.
Table 2: Select descriptive statistics and mean % poverty, the Canadian-born and the foreign-born, aged 70 and older by immigrant status and sex
a Non-family households
C/QPP = Canada or Quebec Pension Plan; OAS/GIS = Old Age Security/Guaranteed Income Supplements
Source: Authors’ calculation from the 2006 Census PUMF of individuals
Such differences between old and new waves of the foreign-born elderly might have reflected differences in their length of residency in Canada. On the one hand, the majority of old-wave seniors had a long period of residence in Canada (e.g., more than 40 years) in comparison to new-wave seniors, who were more likely to have arrived in recent years. The old-wave elders were more likely to have accumulated years of residency and labor market experience in Canada and, hence, were more likely to be eligible for retirement pensions. On the other hand, new-wave elders were more likely to have a shorter period of residence in Canada, and their limited host country labor market experience might have led to a lower probability of receiving retirement incomes.
Gender differences in select characteristics also existed among the elderly population of interest. Overall, elderly women were less likely than men to live only with their spouses, or to receive C/QPPs or retirement pensions (Table 2, columns 1, 3, and 5). In addition, these characteristics varied by immigrant status; more recent arrivals were less likely to receive OAS/GIS, C/QPPs, or retirement pensions, and to live independently (columns 5 and 6).
What can be said about the variations in the select characteristics in poverty rates, defined as the percentages living in households where total annual incomes were below the Statistics Canada LICOs? Table 2 shows these percentages for the Canadian-born and old and new waves of immigrants separately for women and men (columns 7 to 12). In general, the poverty rates of elderly women were substantially higher than those of their male counterparts across all three immigrant status groups (columns 7, 9, and 11 versus columns 8, 10, and 12). The lack of C/QPPs and retirement pensions was associated with the high incidence of poverty (approximately 30%), while association between the receipt of the OAS/GIS and poverty levels was less clear, particularly for women. Moreover, a high concentration of low income was observed among the elderly living alone. In particular, the percentage of unattached elderly women (living alone or with non-relatives) living in low-income households was exceptionally higher than that of the elderly women who had other living arrangements.
Multivariate Analyses
We used logistic regression analysis to examine immigrant and ethnic variations in poverty levels of the elderly after statistically adjusting for group differences in demographic and socioeconomic characteristics. For the convenience of interpretation of results, we converted logits into odds ratios and report odds ratios only. The odds ratios greater than one indicate a higher probability of poverty than the reference group, whereas the odds lower than one means a lower probability of poverty. For reasons of parsimony, the analytical results are presented only for the immigrant-ethnic groups; the full regressions, including logits for the control variables, are available upon request.
In this analysis, we compared immigrant-ethno-racial origin groups with their Canadian-born counterparts of British origin (English, Irish, Scottish, and other British origins). From the mid-1600s on, the British and French were the two major European settler groups in Canada, with the British consolidating economic and cultural control after the defeat of the French in Quebec City during the Seven Years War (1759). Because of its historical and contemporary dominance as a privileged population, we used the British origin group as the reference group in the logistic regression analysis (Nakhaie, Reference Nakhaie2006; Porter, Reference Porter1965).
Most immigrant groups as well as the Canadian-born of non-British origins have significantly higher probabilities of living in low-income households than do the Canadian-born of British origins. This is apparent from the odds ratios that are greater than one in Model 1 in Table 3 (column 1 for women and column 3 for men). Our logistic regression analysis confirms that such poverty gaps are explained largely by demographic and socioeconomic differences between the Canadian-born British origin group and other ethnic-immigrant groups. Comparisons of unadjusted (columns 1 and 3) and adjusted (columns 2 and 4) models in Table 3 show that once distributional differences in demographic and socio-economic characteristics (including retirement incomes and living arrangements) are taken into account, the (logged) likelihood of experiencing low income relative to the Canadian-born of British origins declines; in many instances, the differences are not statistically significant.
Table 3: Odds ratios of being in low income, for women and men, age 70 and older by ethnic-immigrant groups
a Controlled for: age, education, language proficiency, place of residence, year last worked, retirement incomes, and living arrangements
*p < .05; **p < .01; ***p < .001
df = degree of freedom
ns = not significant at p = .05 level; significance levels are based on logits
rg = reference group
Source: Statistics Canada, 2006 Census PUMF of individuals
Nonetheless, two patterns remain after taking group compositional differences into account. First, diminished immigrant-ethnic variation in poverty is more pronounced among Canadian-born and immigrant (from old waves) men than their female counterparts. For men, poverty gaps of all Canadian-born groups and old waves of immigrant groups in comparison to the reference group become non-significant when their socio-economic and demographic characteristics are considered. This implies that higher poverty rates for men in these groups reflect their observed compositional differences more than is the case for women. A second pattern of note is that, for both women and men, elders of new waves of immigration maintain significantly higher propensities to live in low-income households even after controlling for demographic and socioeconomic characteristics. In particular, Jamaican, other West Asian, and East Indian immigrant groups have significantly high poverty levels.
Although Table 3 suggests group-specific demographic and socioeconomic characteristics are factors in explaining immigrant-ethno-racial variation in elder poverty, these analytical results do not demonstrate how state support and family support affect the varied poverty levels among the 31 ethnic-immigrant groups of interest. To clarify the impacts of state support and family support separately, we calculated probabilities of being in poverty using two scenarios. First, we obtained probabilities of poverty for the ethnic-immigrant groups of interest assuming that they had the same state support (i.e., a higher propensity to receive OAS/GIS, C/QPPs, retirement pensions or others, and lower propensity to receive income from other government sources) as the Canadian-born of British origins. For other demographic and socioeconomic characteristics including living arrangements, we used proportional distributions specific to each ethnic-immigrant group. In our second scenario, we computed hypothetical probabilities of living in low-income households, assuming that each group of interest had the same living arrangement characteristics as the Canadian-born of British origins – a tendency towards independent living suggested a lesser degree of financial support from kin. As in the first hypothetical scenario, we used group-specific proportional distributions of other demographic and socioeconomic characteristics (including retirement incomes).
As the percentages in column 2 in Table 4 and 5 suggest, the poverty levels of most groups of the elderly would be lower if they received the same retirement incomes as the Canadian-born of British origins.Footnote 8 In particular, the poverty levels of new immigrant groups, many of whom have racialized ethnic origins, would be much lower both in absolute and relative terms. Overall, the poverty levels of elderly men of new waves of immigration are lower than those of their female counterparts, but they too face sizeable inequalities in the receipt of retirement incomes.
Table 4: Percent reductionFootnote a in the poverty gap between the Canadian-born British and other ethnic-immigrant groups, for women, age 70 and olderFootnote *
a Calculation based on the logits in Model 2, Table 3 (column 2)
(3) = [(2) − (1)]/(1) * 100; (4) = (2) − (1)
(6) = [(5) − (1)]/(1) * 100; (7) = (5) − (1)
CB = Canadian-born
rg = reference group
* Readers may not obtain the same percentage changes shown in columns 3 and 6 if they attempt calculation using the figures in columns 1, 2, and 5 and formulae provided earlier. This is purely due to rounding. For details, please refer to Note 8.
Source: Statistics Canada, 2006 Census PUMF of Individuals
Table 5: Percent reductionFootnote a in the poverty gap between the Canadian-born British and other ethnic-immigrant groups, for men, age 70 and olderFootnote *
a Calculation based on the logits in Model 2, Table 3 (column 4).
(3) = [(2) − (1)]/(1) * 100; (4) = (2) − (1)
(6) = [(5) − (1)]/(1) * 100; (7) = (5) − (1)
CB = Canadian-born
rg = reference group
* Readers may not obtain the same percentage changes shown in columns 3 and 6 if they attempt calculation using the figures in columns 1, 2, and 5 and formulae provided.
This is purely due to rounding. For details, please refer to Note 8.
Source: Statistics Canada, 2006 Census PUMF of Individuals
The spike in poverty rates among elderly immigrants from new waves is striking: poverty levels would increase by 70 to 430 per cent (in percentage change) if new-wave immigrants had the same living arrangements distribution as the Canadian-born of British origins (i.e., a higher propensity towards independent living; see Tables 4 and 5, column 6). The findings suggest that these elderly immigrant groups are protected from poverty because of their living arrangements which include a higher propensity to live with kin. In particular, because of their living arrangements, women of these groups are far less likely to be living in households that are below the LICOs; without these living arrangements, the incidence of poverty would increase by 70 to 430 per cent (in percentage change) for these women (Table 4, column 6). It is also of note that immigrants of Portuguese, Italian, and Greek origins (some of the old-wave groups known for their tendencies to form extended families) would be less well off if they had the same living arrangement characteristics as the Canadian-born British.
Since both state support and family support demonstrably influence the poverty levels of elderly immigrants, we asked which factor had the greater influence on the ethnic-immigrant variations in poverty levels. The proportional reductions in the –2 log likelihood from the baseline model (Table 3) to sequential models (in which either retirement incomes or living arrangements are added to the baseline model) suggest that state support and family support equally influence ethnic-immigrant variations in poverty for women, whereas family support has lesser impact than state support on the poverty levels of men.Footnote 9 The relatively large effect of family support on women reinforces the argument that the poverty of elderly women can be explained by variations in living arrangements which, in turn, reflect differences in economic welfare between elderly women living independently (e.g., living alone) and those living with kin (e.g., living with children).
Conclusion
In this article, we have examined the impacts of state support and family assistance on the poverty levels of immigrant and Canadian-born elders, paying close attention to gender and ethno-racial variations. Results of our logistic regression analyses of the 2006 Census PUMF show that poverty levels among elderly immigrants – particularly those from new waves of immigration – are, on average, higher than those of the mainstream Canadian-born of British descent. This finding is partly explained by the reduced sources of retirement incomes other than OAS/GIS. Our results also demonstrate that the new waves of immigrants with racial labels are protected from poverty to a substantial degree because of their living arrangements (i.e., propensity towards co-residence with kin) even though their poverty levels are high relative to Canadian-born seniors of British origins. In particular, elderly women of Filipino, East Indian, and other South Asian origins benefit from co-residence with kin. Furthermore, we find that while state and family support equally influence the poverty of elderly men, family support explains a greater proportion of the variation in the poverty of elderly women.
This study makes two contributions to the literature on aging and the literature on immigrant economic incorporation. First, it extends the literature on aging by considering simultaneously the impacts of state and family support on the economic well-being of the elderly. Although the economic impact of either state support or family support among recent immigrants has been examined in previous research (Gazso, Reference Gazso2005; Lewin & Stier, Reference Lewin and Stier2003; Rendall & Speare, Reference Rendall and Speare1995), these supports are seldom considered together in multivariate analyses of the economic well-being of older people. By contrast, our study has considered the impact of both, calculating the hypothetical percentages of poverty among the immigrant and non-immigrant elderly of various ethnic origins and using the distributions of living arrangements and retirement income for Canadian-born seniors of British origins as the reference group. A second contribution is that our research adds to the literature on immigrant economic incorporation by providing further evidence of challenges faced by new waves of immigrants with racialized ethnic origins. Our results illuminate substantially high levels of poverty among immigrant elders belonging to this group, set within the context of an overall decline in elder poverty in Canada over the past several decades.
Although our study improves upon the literature in these regards, limitations and unanswered questions remain. Accordingly, our analysis likely underestimates the effect of state support on the poverty of the elderly in Canada as a result of our analytical strategy. Our analysis considered disparities in the access to retirement income sources by nativity and ethno-racial origins, yet we did not consider disparities in the amount of C/QPP and private pensions linked to the disparities in employment earnings while the elderly under analysis were in the labor force. Given the known inequalities in the labor market by race, ethnicity, and nativity, we may likely observe internal differences in the amount of C/QPP and private pensions among those elders receiving these retirement income sources. This may further explain the gaps in poverty levels between the Canadian-born British group and others especially the new-wave immigrant groups, who are more likely to have a shorter period of residence in Canada and to belong to ethno-racial minorities.
Moreover, although our analysis shows that co-residence with kin has a positive impact on the poverty alleviation of the elderly, drawbacks to co-residence with kin also need to be acknowledged. As suggested in previous studies, living with adult children and/or other relatives may have a negative impact on the mental health of the elderly (McLaren, Reference McLaren2006; Wilmoth & Chen, Reference Wilmoth and Chen2003). For instance, seniors living with kin may develop a feeling of isolation if they spend substantial amounts of time on household work (e.g., caring for grandchildren) and consequently have little social interaction outside of home. In particular, elderly women may be at a greater risk of experiencing such isolation, considering gender inequalities in families in which women are – more often than men – deemed as the primary caregivers at home. Moreover, elderly immigrants may face additional drawbacks to co-residence with kin due to conflicts with their family members that can arise from intergenerational differences in the integration into the host society. For elderly immigrants, widening gaps in cultural identities and proficiency in the host country language from their co-residing kin in younger age may isolate them psychologically, and thus negatively affect their mental health.
To conclude, our results, especially the illumination of high poverty levels among minority seniors from new waves of immigration, have implications for public policy and policy debate on the economic well-being of immigrant seniors. One policy action that could improve the economic well-being of immigrant seniors is the implementation of international agreements on retirement incomes. In Canada, the International Social Security Agreement (ISSA) is particularly relevant to the issue of poverty among elderly immigrants. This agreement stipulates that immigrants may use pensions from their countries of origin and may use periods of residency there to qualify sooner for Canadian old age security. The ISSA ameliorates the poverty of recently arrived immigrant seniors who do not meet minimum qualifying requirements for Canadian retirement incomes such as OAS or C/QPP (Elgersma, Reference Elgersma2007). However, not all recent immigrants benefit. To date, the Canadian agreement includes only 50 countries, largely European (e.g., Germany, Italy, Portugal), a concentration that reflects trends in immigration when the ISSAs evolved, namely the late 1970s and 1980s when Europe still dominated Canada’s immigrant source region. Seniors of the new waves of immigration (largely from Asia and the Middle East) are less likely to benefit unless agreements with the newer sending countries are struck. In addition, these new-wave origin countries have less-developed welfare states in general. Therefore, these countries currently may not have developed income programs for the elderly comparable to Canada’s retirement income systems for reaching the ISSAs in the near future (Durst, Reference Durst2005).
A second Canadian policy initiative relevant to the poverty of seniors in new waves of immigration would be to amend the Old Age Security Act.Footnote 10 This was attempted recently but without success. Bill 362: An Act to amend the Old Age Security Act – a private member’s bill proposed by Colleen Beaumier (Liberal MP, Brampton West) – was tabled in October 2006. It proposed to reduce the residency requirement for OAS from 10 years to 3 years. Although this bill was supported by some ethnic and racial minority communities because of its potential to improve the economic well-being of their elderly populations, critics argued that current residency requirements do not violate rights or constitute discrimination, citing the Charter of Rights to back up their arguments (African Canadian Social Development Council, 2010; Chinese Canadian National Council Toronto Chapter, 2006; Elgersma, Reference Elgersma2007).
Bill 362 was also opposed by the Conservative members of the Standing Committee on Human Resources, Social Development and the Status of Persons with Disabilities in May 2008. As of May 16, 2008, it was awaiting the third reading required under Canada’s parliamentary system to receive Royal Assent and become legislation (Parliament of Canada, no date) although some thought it not likely to proceed to a vote at Third Reading because it was a private member’s bill (Beaumier, Reference Beaumier2008). On September 7, 2008, Prime Minister Harper received permission from the Governor General to dissolve Parliament and hold a general election on October 14, 2008, with the result that all pending bills were terminated. Beaumier did not run in that election, and proposals to reduce the OAS residency requirements have not been re-introduced since then. To date, fiscal policy initiatives targeted at the newest waves of elderly immigrants appear to be at a standstill, thus perpetuating the importance of family living arrangements in providing financial safety nets for these and other elderly immigrants.
