Social Assistance Receipt Among Refugee Claimants in Canada: Evidence from Linked Administrative Data Files
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.
by Yuqian Lu, Marc Frenette, and Grant Schellenberg, Social Analysis and Modelling Division, Statistics Canada
- Executive summary
- 1 Introduction
- 2 Data and methodology
- 3 Results: Receipt of social assistance income
- 4 The individual characteristics of refugee claimants and social assistance receipt
- 5 Conclusion
- 6 Appendix Tables
Refugee claimants are an important part of the non-permanent resident population of Canada. Canada granted permanent residency to approximately 12,000 to 16,000 refugees every year during the latter part of the 2000s, and approximately 115,000 to 130,000 refugee claimants were residing in Canada at some point every year over that period. Despite the volume of refugee claimants, very little information on their economic characteristics has been available to date. This report draws on new linked administrative data files to provide information on the receipt of social assistance (SA) among this population. The study was successful in linking approximately three-quarters of all refugee claimants to administrative files containing information on the annual receipt of SA. Of these linked claimants, approximately 80% rely on SA shortly after opening their refugee claim. Among those whose claims were still open after four years, between 25% and 40% were collecting SA, well above the overall Canadian rate of about 8%. The variation in SA rates tends to be strongest according to province of residence and country of citizenship. Some differences exist by family type, age of oldest family member, and presence of a work permit within the family, but they are not nearly as pronounced. Between 85% and 90% of refugee claimants under the age of 19 are in a family that receives SA shortly after making a refugee claim. Overall, between $10 billion and $13 billion were disbursed every year to all SA recipients in Canada. Given their relatively small size as a group, the dollar amount of SA paid to refugee claimant families amounted to between 1.9% and 4.4% of that total, depending on the year and on the treatment of unlinked cases.
Keywords: Refugee claimants, social assistance
Prior to the December 2012 reforms to the in-Canada refugee determination system, it was possible for some refugee claimants to reside in Canada for several years before the final adjudication of their claim. Their labour market activities, their receipt of publicly funded benefits, and their well-being during that period are relevant to public policy. Nonetheless, very little information on the economic characteristics of refugee claimants has been available to date.
This report begins to address this data gap. The Refugee Claimant Continuum Database from Citizenship and Immigration Canada and administrative tax files were combined into a new linked dataset capable of supporting research on refugee claimants. The focus in this report is on the receipt of social assistance (SA) among refugee claimants who initiated their claim between 1999 and 2011.
Among refugee claimants of all ages, 74% to 78% were linked to administrative tax data over the period from 2002 to 2011; linkage rates were somewhat lower between 1999 and 2001, at 67% to 71%. Alternative measures of SA receipt were estimated to assess the impact of unlinked cases.
In the year following the start of their refugee claim (i.e., in year t+1), the estimated rate of SA receipt varied somewhat depending on the treatment of unlinked cases. The lower-bound estimate, which assumed that all unlinked cases did not receive SA was about 65%. The upper-bound estimate assumed that all unlinked cases received SA, and was about 85%. The middle estimate, which simply excluded all unlinked cases, was close to the upper-bound estimate (about 80%).
Focusing on the middle estimate, the receipt of SA in year t+1 among the 2005-to-2010 claimant cohorts generally ranged between 80% and 90% across family types, with rates highest among lone mothers and couples with more than two children. Similarly, the incidence of SA receipt generally ranged from about 80% to 90% across families in which the oldest member was between 19 to 24 and 55 to 64 years of age. Across provinces, the incidence of SA receipt in year t+1 was generally highest in Quebec, at over 85%, and lowest in Alberta, at under 60%.
SA receipt varied considerably across country of citizenship. Refugee claimants from countries such as Afghanistan, Colombia, the Democratic Republic of Congo, Eritrea, and Somalia all had relatively high SA rates (close to or above 90%) throughout most of the study period, while rates were lower among refugee claimants from Bangladesh, Haiti, India, and Jamaica (generally below 80%).
The rates of SA receipt tended to decline sharply in the years following the start of the refugee claim. Between years t+1 and t+2, rates fell by about 20 percentage points among most claimant cohorts, declining a further 15 percentage points between t+2 and t+3, and 10 percentage points between t+3 and t+4. By t+4, between 25% and 40% of refugee claimants received SA. However, it is important to recall that these figures pertain to the diminishing group of refugee claimants whose claims remained open up to that year. These figures are also well above the Canadian average of about 8%.
Among refugee claimant families that received SA in year t+1, the average total family income typically ranged from about $19,000 to $22,000, with SA benefits accounting for $8,000 to $11,000—or about 40% to 48%—of that total.
In aggregate terms, SA income paid to all recipients in Canada totaled $10 billion to $13 billion in most years. Given their relatively small size as a group, the dollar amount of SA paid to refugee claimant families amounted to between 1.9% and 4.4% of that total, depending on the year and on the treatment of unlinked cases.
Canada granted permanent residency to approximately 12,000 to 16,000 refugees every year through the latter part of the 2000s. However, given ongoing inflows of refugee claimants and the time required to adjudicate claims, approximately 115,000 to 130,000 refugee claimants were residing in Canada at some point every year over that period.
Currently, information on the financial characteristics of refugee claimants is scarce. Two data files often used for immigration research—the Longitudinal Immigration Database (IMDB) and the Longitudinal Administrative Databank (LAD)—do not allow researchers to examine any income information for refugee claimants who have had their claims denied. This data gap is significant, considering that refugee claimants may reside in Canada for several years before the final adjudication of their claim, and considering that their labour market activities, their receipt of publicly funded benefits, and their well-being during this period are relevant to public policy.Note 1,Note 2,Note 3
Statistics Canada and Citizenship and Immigration Canada (CIC) are working collaboratively to address this data gap. Information from both immigration-based and taxation-based data files have been combined into a new linked dataset capable of supporting analysis of the economic characteristics of refugee claimants: the Refugee Claimant Database (RCD). The first part of this report describes the methodology and the contents of this data file.
Although the RCD is capable of supporting research on many aspects of the characteristics of refugee claimants and of their experiences in Canada, the focus in this report is on the receipt of social assistance (SA) income. SA programs operated at municipal, provincial, and federal levels of government are designed to provide income to meet the costs of basic requirements for individuals or families when all other financial resources have been exhausted. Given the traumatic and disruptive events recently faced by many refugee claimants and their weak ties to the Canadian labour market, one might well expect many of them to rely on SA as a main source of income.Note 4 The extent to which this is the case is the main focus of this report.
The rest of the report consists of three sections. In Section 2, a detailed discussion of the design and construction of the analytical data file is provided. The target population, source data files, and linkage strategies and rates are discussed in detail. In Section 3, analytical results are presented, with a focus on the incidence of SA receipt among refugee claimants through the 2000s. In the final section, conclusions and directions for further analytical work are presented.
The target population consists of refugee claimants. The Temporary Resident database from CIC contains information on all non-permanent residents in Canada, including refugee claimants. Specifically, refugee claimants are individuals who...
- “...request refugee protection upon or after arrival in Canada. A refugee claimant receives Canada’s protection when he or she is found to be a Convention refugee as defined by the United Nations 1951 Geneva Convention Relating to the Status of Refugees and its 1967 protocol, or when found to be a person needing protection based on risk to life, risk of cruel and unusual treatment or punishment, or danger of torture as defined in the Convention Against Torture. A refugee claimant whose claim is accepted may make an application in Canada for permanent residence. The application may include family members in Canada and abroad.” (CIC 2013).
Five input data sets were used to construct the analytical file used for the analysis in this report.
Refugee Claimant Continuum Database (RCCD): The RCCD contains socio-demographic and administrative information on refugee claimants. Refugee claimants were identified on the basis of their “yearly primary status” (as defined by CIC). The reference period for their claim was determined using the dates on which their claim was opened and closed (if applicable). A family identification number on the RCCD facilitated identification of all members of refugee claimant families (including minor children) in the year the refugee claim was initiated.
Linkage Control File (LCF): The Temporary Resident database from CIC does not contain the Social Insurance Numbers (SINs) of temporary residents, and therefore, cannot be deterministically linked to taxation-based administrative files. To bridge this gap, the LCF from the Household Survey Methods Division at Statistics Canada was used (more on this below).
T1 Family File (T1FF): The T1FF is a census of T1 personal income tax forms, with steps taken to construct family-level information for each tax year. The T1FF family concept is similar to the census family concept (parent[s] and co-resident children). The file includes demographic, earnings and income information, including a unique field for SA income. In families with two spouses present, the one with the higher net income must report the SA income on the T1 tax return. However, most individuals in families that receive SA income can be identified using the T1FF family-level identifier. If this was not possible, the family identification number from the RCCD file was used.
T5007 file: The T5007 file is derived from the T5007 Statement of Benefits forms issued by the provincial, territorial or municipal social service agencies making the payments. It is a summary of worker’s compensation benefits and SA payments, or provincial or territorial supplements; for this project, only the latter is of interest. These are primarily basic SA (support payments that are not targeted at particular individuals, other than those in need) and support payments (supplements) for elderly and disabled individuals in need. Such payments are grouped together on the T5007 file, and are referred to as “social assistance” (SA) in this report. The individual who receives a T5007 form for SA reporting is the “principal claimant.” This person can be any adult member of the family, not necessarily the same person who reports family SA income on the T1 form. Because the T5007 file contains no family information, it is not possible to construct family SA income from this file. However, once the T5007 file is linked to the T1FF, the SA income reported on the T5007 file can be aggregated at the family level.
Data quality assessment by Statistics Canada indicates that total aggregate SA income calculated from the T5007 file is somewhat higher than that calculated from the T1FF, with most of the difference resulting from undercoverage of SA recipients on the T1FF. This undercoverage has little impact on findings pertaining to SA use by the general population. However, undercoverage of T5007 SA recipients on the T1FF may have a larger impact on results for refugee claimants if this population (particularly refugee claimants who do not subsequently land) are disproportionately unlikely to file a T1 tax return.
T1 Historical (T1H) File: The T1FF is constructed using the T1 Personal Master Files compiled by the Canada Revenue Agency (CRA) 12 to 18 months after the end of each tax year. Therefore, it does not include the T1 returns of individuals who filed their taxes at a later date. The omission is relevant to this study, because refugee claimants may be more likely than the general population to file their T1 tax return late, owing to their (i) recent arrival in Canada, (ii) potential lack of knowledge of the Canadian tax system, (iii) low taxable income, and (iv) settlement challenges associated with involuntary dislocation from their country of origin. The T1 Historical Personal Master file is compiled by CRA approximately four years after the end of each tax year, and therefore, includes “late filers” as well as re-assessed T1 returns. T1H data are used to test the robustness of the main results in this paper by addressing undercoverage due to late filing.
Data linkage methodology and linkage rates
Several steps were taken to create the Refugee Claimant Database (RCD). First, seven variables from the RCCD were used to probabilistically match individuals to the Linkage Control File (LCF). These variables were: first and last name, date of birth, gender, postal code, claim date and family indicator. The SINs of linked refugee claimants were then obtained from the LCF, yielding a linkage key containing both the CIC Client Identification Number and the SIN. Next, a deterministic linkage based on SIN was established between these refugee claimants and the T1FF, the T5007 file, and (to assess the impact of late filing among refugee claimants) the T1H.
Since 2002, this process yielded a successful linkage rate between the RCCD and taxation-based administrative files of about 59% to 62% (the bas showing the individual linkage rate, i.e., the dark blue bars in Chart 1). Prior to 2002, linkage rates ranged from 53% to 57%. However, it is important to note that children and youth are included in this rate even though most of them do not file tax returns or receive government forms such as the T5007 Statement of Benefits. Consequently, they cannot be directly linked.
In studies of labour market activities, sample populations are typically restricted to working-age individuals, and the unlinked records of children and youth are not a concern. However, SA is intended to provide financial support to all members of recipient families and it is important to consider refugee claimants of all ages. To do so, family-level identifiers were used to capture non-filing individuals in the RCD. Refugee claimants of all ages were first grouped into family units using the T1FF family identifier. In cases where this strategy was not successful, refugee claimants of all ages were grouped into family units using CIC’s family identification number.Note 5 Individuals identified in these ways were then coded as residing in refugee families that either did or did not receive SA. The use of family identifiers yielded linkage rates for refugee claimants of all ages that ranged from 74% to 78% between 2002 and 2011, and from 67% to 71% between 1999 and 2001 (the bars showing the family linkage rates, i.e., the light blue bars in Chart 1).
With the information available on the RCCD, it is not possible to identify refugee claimants who left Canada prior to the adjudication of their claim without informing CIC. These ‘departed’ refugee claimants remain on the RCCD until their claim is identified as abandoned and then closed. This will reduce the linkage rates between the TR and administrative files since departed refugee claimants would be included in the denominator of linkage rate calculations (i.e., in the claims identified as active) but not in the numerator since individuals were not present in Canada to file a T1 tax return or to receive a T4 Slip. The magnitude of this effect cannot be estimated.
Refugee claimants who could not be linked into the analytical file differed in systematic ways from those who were linked. Table 1 shows the mean family characteristics (in year t+1) associated with being linked or unlinked. The focus is on family-level characteristics in this table and throughout most of the report, as SA eligibility is determined at the family level. Note that all characteristics are taken from the RCCD in the year the refugee claim was initially made (i.e., in year t). The one exception is province of residence, given the possibility that refugee claimants may be highly mobile after arriving in Canada. Province of residence in year t+1 is obtained from the T1FF, and if that information is not available, from the T5007 file. In unlinked cases, the province of residence is taken from the RCCD in year t.
The likelihood of remaining unlinked in the RCD was highest among refugee claimants who were single individuals, residing in ‘young’ families, and who resided in families in which no one received a work permit. More specifically, refugee claimants residing as single individuals accounted for 54% of all linked observations, but for 76% of all unlinked observations. Unlinked individuals were also more likely to be in a family whose oldest member was under 19 years of age (12%), while this was the case for less than 1% of linked individuals.Note 6 And while refugee claimants residing in a family in which no one held a work permit accounted for 7% of linked cases, they accounted for 54% of unlinked cases. The geographic distribution of linked and unlinked cases was roughly the same, with only small differences existing between the two. The same can be said about the distribution of country of citizenship and the year in question.
The differences that exist in family type, age of oldest family member, and the presence of a work permit in the family may matter for SA receipt (to the extent that these factors are correlated with SA). For this reason, the treatment of unlinked cases in the analysis is of critical importance—a topic that is addressed next.
Defining social assistance and its incidence of receipt
For the years during which their refugee claim was ongoing, refugee claimants were identified as being in receipt of SA income if (i) they or anyone in their family reported SA income on a T1 Personal Tax Return and/or (ii) they or anyone in their family were issued a T5007 form indicating receipt of income from SA or provincial supplements.
Refugee claimants who remain unlinked cannot be ascribed a value of SA income. Given that approximately one-quarter of refugee claimants are unlinked, the treatment of these unlinked cases has a significant impact on the estimation of SA receipt among refugee claimants. To provide a clear assessment of this impact, alternative measures of SA receipt are provided:
Lower-bound estimate: All refugee claimants (i.e., linked and unlinked) are included in the denominator of the SA receipt calculation, but only linked refugee claimants who received SA are included in the numerator. In short, it is assumed that none of the unlinked refugee claimants received SA.
Upper-bound estimate: All refugee claimants (i.e., linked and unlinked) are included in the denominator of the SA receipt calculation, and both linked refugee claimants who received SA and unlinked refugee claimants are included in the numerator. In short, it is assumed that all unlinked refugee claimants received SA.
Middle estimate: Only linked refugee claimants are included in the SA receipt calculation; unlinked cases are excluded from both the numerator and denominator. In short, SA receipt is calculated only for refugee claimants for whom information is available.
While estimates of SA rates will be documented using all three definitions, subsequent analysis of the covariates associated with SA receipt will mainly be documented using the middle estimate. This approach is agnostic in nature since the results pertain only to cases that are linked and no assumptions are made regarding the SA receipt of unlinked cases. However, some may assume, de facto, that the receipt of SA observed among linked cases is identical to that among unlinked cases. This may not be the case if (i) the composition of refugee claimants who are linked is different than the composition of refugee claimants who remain unlinked—a fact documented in Table 1—and (ii) if the likelihood of SA receipt among unlinked cases is different than that of linked cases conditional on having similar characteristics.
To probe this issue further, the following steps were taken. First, a regression model was run in which an SA dummy variable was regressed on the family characteristics identified in Table 1 for linked cases only. It was then assumed that the statistical relationships between SA receipt and these variables were the same for linked and unlinked cases. The regression coefficients were then used to predict SA receipt among the full population of refugee claimants—including both linked and unlinked observations. Despite the compositional differences between linked and unlinked cases, this exercise yielded predicted SA rates very similar to those observed among the linked population alone (i.e., the middle estimate defined above).Note 7 This is because some of the characteristics disproportionately found among unlinked cases were associated with a lower probability of receiving SA (e.g., being in a young family), while other characteristics disproportionately found among unlinked cases were associated with a higher probability of receiving SA (e.g., being in a family with no work permit).Note 8 Such factors offset each other, yielding little difference between the predicted rate among the total population and the observed rate among the linked sample.
This assessment suggests that the compositional differences between linked and unlinked cases are not likely to lead to large differences in overall SA rates between the two groups. Thus, under the assumption that the relationships between SA receipt and the family characteristics identified in Table 1 hold for both linked and unlinked cases, the middle estimate likely serves as a reasonable proxy for SA receipt among all refugee claimants. That being said, differences in linkage rates across socio-demographic characteristics mean that estimates of SA receipt for specific groups of refugee claimants (e.g., young families, families without a work permit) will be more sensitive to how unlinked cases are treated, yielding greater variability in estimates of SA rates for subgroups. This is addressed in more detail below.
Claim cohorts, years since claim, and the duration of refugee claims
The analysis presented below is predominantly annual in nature. While specific start and termination dates are available for each refugee claim on the RCCD, taxation-based administrative data are annual. Hence, while it is possible to determine whether or not a refugee claimant resided in a family that received SA income in a given year, it cannot be determined in which months of that year the income was received.
Similarly, refugee claims are largely treated as annual data. The start date of each claim is used to identify the year in which the refugee claimant arrived, and the termination date of the claim (if applicable), is used to identify the year in which the claim was closed. An individual whose refugee claim was initiated in September 2007 and closed in February 2009 would be identified as a refugee claimant in 2007, 2008, and 2009. The individual would be out-of-scope in 2010 on the grounds that he or she were no longer a refugee claimant. Information on the status of refugee claims is not available during the intervening years. The term “open claims” is used in this report to refer to claims present on the file, although in some cases an Immigrant Refugee Board decision may have already been made on these claims.
The study includes refugee claimants whose claims were initiated between 1999 and 2011. ‘Claim cohorts’ are defined in terms of the first year of the refugee claim (i.e., the 1999 arrival cohort). In addition, the first year of the refugee claim is identified as year ‘t’, while subsequent years after arrival are identified as years ‘t+1’, ‘t+2’, and so on.
It is important to note that the size of each refugee claimant cohort declines over time as claims are adjudicated and closed. As shown in Table 2, for example, 28,322 refugee claims were started in 2007 and the vast majority of these were still active in t+1 and, to a lesser extent, in t+2. But by t+4 most of these refugee claims had been adjudicated and closed, with 11,598 claims still active. One implication is that the incidence of SA receipt over time is calculated for a diminishing number of refugee claimants. This proportion varies across cohorts, with about 26% of refugee claimants in the 2004 and 2005 cohorts still present in t+4 compared with just over 40% of refugee claimants in the 2001 and 2007 arrival cohorts. Given this variability, much of the analytical focus is on SA receipt in t+1 when most refugee claims were still active.Note 9 Some analyses look at later years, but these are most relevant after adjusting for differences in the socio-economic characteristics of refugee claimants with claims ongoing after several years.
Compositional characteristics of refugee claimants in the analytical file
Refugee claimant families come to Canada with many different characteristics and these may be associated with SA receipt. These were shown in Table 1 for linked cases across all years of analysis; however, the goal of that table was to highlight differences between linked and unlinked cases. In Table 3, mean family characteristics of all linked cases are shown across cohorts in year ‘t+1’. Depending on the cohort, between 47% and 62% of refugee claimants are single (unattached) individuals. In most years, about 15% of refugee claimants reside in lone-mother families and about 2% reside in lone-father families. About 11% to 18% of refugee claimants reside in families characterized by couples with two or more children. In most years, about two-thirds of refugee claimants reside in a family where the oldest member is aged 25 to 44, and over 80% reside in a family where the oldest member is aged 25 to 54. More than 80% of refugee claimants reside in a family with at least one member holding a work permit, with this share above 95% beginning with the 2003 claim cohort.
Between 82% and 93% of refugee claimants reside in Ontario or Quebec. Of that group, most are in Ontario. For example, among the 2010 cohort of refugee claimants, 67% lived in Ontario in year t+1, while 20% lived in Quebec. The next most common provinces were British Columbia (accounting for 3% to 8%) and Alberta (accounting for 2% to 6%). Taken as a whole, the other provinces and territories usually account for about 2% of all refugee claimants.
Refugee claimants also come from various countries. The 21 countries of citizenship listed in Table 3 collectively account for roughly half of all refugee claimants. Throughout most of the period, the most common countries included the People’s Republic of China, Colombia, Sri Lanka, and (in more recent years) Haiti and Hungary.
The incidence of SA receipt among refugee claimants in year t+1 are shown in Chart 2 using the three methods discussed above for dealing with unlinked cases. Overall, the exclusion of unlinked cases yields estimates that are closer to the upper bound estimates.Note 10 Also, in all cases, SA receipt was lowest among the 2002 and 2003 refugee claimant cohorts, and highest among cohorts who initiated their claim in 2007, 2008, or 2009 (and thus, may have been affected by the economic downturn that started in late 2008). Specifically, the lower bound estimates—which assume that all unlinked refugee claimants did not receive SA—were somewhat lower among the 2002 and 2003 refugee claimant cohorts (59% and 55%, respectively) and somewhat higher among the 2007 and 2008 cohorts (at 71% and 70%, respectively). The upper bound estimates—which assume that all unlinked cases did receive SA—were again somewhat lower among the 2003 cohort (79%) compared to latter cohorts (e.g., 86% among the 2007 cohort). If unlinked cases are simply dropped from the calculation (i.e., the middle estimates), the incidence of SA receipt was again lower among the 2003 cohort (72%) compared to latter cohorts (e.g., 84% among the 2007 and 2008 cohorts).
Characteristics associated with the likelihood of refugee claimants receiving social assistance
The family characteristics shown in Table 3 may be associated with SA receipt. To begin examining this relationship, the SA rates in year t+1 are calculated for individuals by the family-level characteristics in Table 4. Selected results are shown graphically in Charts 3 to 7. These results are based on the middle estimates, with unlinked cases excluded from estimates of SA receipt. The total sample counts are shown in Appendix Table 1. The final column in Table 4 shows weighted average SA rates at t+1 for the six claimant cohorts that started in the years 2005 to 2010 and is provided as a summary measure.
The results show that the incidence of SA receipt varies to some degree across each of the family characteristics. Specifically, SA rates are generally lowest among single individuals (and to a lesser extent, childless couples) and generally highest among lone mothers and couples with more than two children. However, the magnitude of the difference across family types is not very large, with a difference of about 15 percentage points in most years across family types with the lowest and highest rates.
Similarly, the variation in SA rates across the age of the oldest family member is quite modest. As shown in Chart 4, the incidence of SA receipt typically varies by less than 5 percentage points between refugee claimants in families whose oldest member is aged 19 to 24 and those in families whose oldest member is aged 45 to 54. The incidence of SA receipt is slightly higher in families whose oldest member is aged 55 or older, but, as noted above, only about 5% of all refugee claimants reside in such families.
Holding a work permit may be expected to be negatively associated with SA receipt since individuals have the option of working. This was certainly the case for the 1999 to 2003 claim cohorts as the incidence of SA receipt among refugee families with a least one member holding a work permit was 8 to 12 percentage points lower than among refugee families in which no one held a work permit. For the 2010 claim cohort, the difference between these groups was 5 percentage points. However, for the 2005 to 2010 cohorts, SA rates were generally about the same for both groups. Over the entire period, SA rates were never below 72% among families with at least one member holding a work permit.
As noted earlier, most refugee claimants reside in one of four provinces: Quebec, Ontario, Alberta, and British Columbia. Of these four, SA rates are highest in Quebec, where between 79% and 91% of refugee claimants receive SA in year t+1. In contrast, generally between one-half and two-thirds of refugee claimants residing in Alberta and between one-half and three-quarters of refugee claimants residing in British Columbia collected SA. Among cohorts arriving before 2007, SA rates were at least 9 percentage points lower in Ontario than in Quebec. Since then, the rates have been quite close in both provinces and, in fact, rates were slightly higher in Ontario than in Quebec among the 2009 and 2010 cohorts.
The country of citizenship of refugee claimants is also strongly associated with SA receipt. While there are variations over time, some trends hold throughout most of the study period. Individuals from countries such as Afghanistan, Colombia, the Democratic Republic of Congo, Eritrea, and Somalia all have relatively high SA rates (close to or above 90%) throughout most of the study period. At the opposite end of the spectrum, SA receipt is relatively low throughout the period for refugee claimants from Bangladesh, Haiti, India, and Jamaica. The rates are particularly low among Jamaican refugee claimants who arrived before 2007, never surpassing 30%. Since then, rates have remained slightly above 40%, which is still the lowest among the countries examined here and only about half of the overall average.
In summary, the variation in SA rates tends to be strongest across the province of residence and the country of citizenship. Some differences exist by family type, age of oldest family member, and presence of a work permit within the family, but it is nowhere nearly as pronounced as what is registered by province of residence and country of citizenship.
The patterns revealed by Table 4 are largely confirmed by regression analysis. In Table 5, results from a linear probability model (ordinary least squares) of SA use are shown, where the covariates include all of the socio-economic factors appearing in Table 4.Note 11,Note 12 In general, the magnitude of the differences tends to be somewhat different once the socio-economic characteristics are taken into account, but the trends are still qualitatively the same. The magnitude is different because many of these characteristics are correlated, and only through a multivariate approach such as regression analysis can the separate (independent) relationships between SA use and each factor be determined.
The predicted difference in SA use between the family type with the highest SA rate (couples with more than two children) and the lowest SA rate (single men) is 15.2 percentage points. The model also suggests that older families (those with someone aged 55 to 64) are more likely to receive SA than younger families. Predicted SA rates are about 6 percentage points lower among families where a work permit is held by a member.
The regression results also suggest substantial differences across the main provinces of residence. Compared to Ontario, predicted SA rates are 13 percentage points higher in Quebec, and about 10 and 22 percentage points lower in British Columbia and Alberta, respectively. The difference between the highest (Quebec) and lowest (Alberta) predicted SA rate is 35 percentage points.
Important differences by country of citizenship are also registered. The reference category is the People’s Republic of China, which has a relatively high rate of SA use. In fact, only refugee claimants from Eritrea and Somalia have higher predicted SA rates. The lowest predicted SA rates are among refugee claimants from Jamaica, followed by refugee claimants from Haiti, India, and Syria. The predicted SA rate among refugee claimants from Jamaica is once again far below all others (53 percentage points below refugee claimants from the People’s Republic of China).
Thus, the regression analysis confirms the finding in Table 4 that the variation in SA rates is larger across the province of residence and the country of citizenship compared to other family characteristics such as the family type, the age of the oldest member, and whether or not a family member holds a work permit.
The incidence of SA receipt was lowest among the 2003 refugee claim cohort (the reference category) as all other cohort coefficients are positive and significant, and highest among the 2007, 2008, and 2009 cohorts—all of whom arrived in Canada around the time of the economic downturn. According to LAD data, SA rates among the general population only rose marginally during the economic downturn—going from 7.4% in 2007, to 7.7% in 2009.
Receipt of social assistance across socio-demographic characteristics—sensitivity to treatment of unlinked cases
The incidences of SA receipt shown above are based on the middle estimate—that is, refugee claimants who could not be linked to taxation-based administrative data were excluded from the calculations. One question this raises is how much the rates of SA receipt vary for specific groups when different methodologies are used to handle unlinked cases. This is documented in Table 6 which shows the incidence of SA receipt for the 2008 claim cohort in year t+1 across selected characteristics using the lower-bound, middle, and upper-bound estimates as described in Section 2.
The incidence of SA receipt varies most across estimates for refugee claimants residing as single individuals. The lower-bound estimate of SA receipt among single men and women is 60% to 63%, but this jumps to 82% to 86% with the upper-bound estimate. This is a range of 22 percentage points across the three approaches. In contrast, the range in the estimates of SA incidence across other family types is 6 to 13 percentage points. This reflects the fact that refugee claimants residing as single individuals account for a relatively large share of unlinked observations.
The same results are evident across age of the older family member. Among refugee claimants residing in families in which the oldest member is between age 35 and 54 the estimated incidence of SA receipt ranges from 75% (lower-bound) to 85% to 86% (upper-bound)—a magnitude of about 10 percentage points. The range among families whose oldest member is between 25 and 34 years of age is 16 percentage points. In contrast, refugee claimants aged 19 to 24 are overrepresented among unlinked observations and the estimated incidence of SA receipt ranges from 62% to 87%—a range of 25 percentage points.
In short, the estimated incidences of SA receipt among refugee claimants residing in families— especially those in which the oldest member is older than age 35—are least sensitive to how unlinked observations are treated. The same can be said of refugee claimants residing in families in which at least one member holds a work permit, and among refugee claimants residing in Quebec, Ontario and Alberta.
Appendix Tables 2 and 3 provide estimates of SA receipt across the same characteristics shown in Table 4 above, but for the lower-bound and upper-bound estimates, respectively.
Social assistance receipt by years since claim
This section examines the trend in SA rates by the number of years since the start of the claim. The focus here is once again on the middle estimate. The SA rates of refugee claimants fall quickly after year t+1. In Chart 8, the rates are shown for each refugee claimant cohort for up to four years following the beginning of the refugee claim. Between years t+1 and t+2, SA rates fall by roughly 20 percentage points. The decline is also large between t+2 and t+3, falling by a further 15 percentage points, on average. And between t+3 and t+4, SA rates fall by about another 10 percentage points. Overall, between t+1 and t+4, SA rates generally fall by more than about half—or by 45 to 50 percentage points. However, it is important to recall that these figures pertain to the diminishing group of refugee claimants whose claims remained open up to the year in question, as shown earlier in Table 2.
The first refugee claimant cohort—those who claimed in 1999—can be followed for 12 years, providing a much longer view of SA receipt. As suggested by Chart 9, the SA rate fell from 82% in year t+1 to 26% by year t+6; it subsequently remained between 21% and 26% through to year t+12. Similar results are obtained for the 2001 and 2003 claim cohorts, who can be followed for up to 8 and 10 years after the start of their claim, respectively. Note that, once again, these figures pertain to the group of refugee claimants whose claim remained open up to the year in question. Of all refugee claimants in the 2001 cohort, 9% were observed in t+8 while less than 3% of claimants in the 1999 cohort were observed in t+12.
The steep decline in SA rates as the refugee claim becomes older may be related, at least in part, to differences in characteristics of individuals whose claim lasted for different lengths of time. To account for this possibility, an ordinary least squares regression of SA rate was estimated for the 1999 claim cohort, with all years since the start of the claim pooled in the same regression.Note 13 This regression accounts for differences in all of the family characteristics examined so far. In Table 7, the coefficients of interest are those associated with the variables denoting the number of years since the start of the claim. The declining SA rates are evident in these results. By year t+12, predicted SA rates are 51.6 percentage points lower than in year t+1 after accounting for changes in the composition of refugee claimants over that period. This difference is only somewhat smaller than the unadjusted results shown in Chart 9 (56 percentage points). The results are similar for other cohorts.
While considerably lower than at the beginning of the refugee claim, SA receipt is still far more common among the refugee claimant population 12 years following their initial claim than among the broad population. To demonstrate this, the overall rate of annual SA receipt among the Canadian population was calculated with the LAD between 1999 and 2011.Note 14 Since 2003, the rate has been fairly steady at around 8%. At no point does the rate surpass 10%, and it is always less than a third of the rate of receipt among refugee claimants from the 1999 cohort, even 12 years after their initial claim (26%).Note 15
The share of total aggregate social assistance paid to refugee claimants
So far, only the incidence of SA receipt has been examined. However, it is worthwhile to consider a broader perspective, and assess how much of the total aggregate SA expenditures disbursed by federal, provincial and municipal government departments and agencies are paid to refugee claimants. Although refugee claimant families are far more likely than the broader Canadian population to receive SA income, it is important to keep in mind that refugee claimants only represent about one-third of 1% of the overall Canadian population.
In Table 8, the total SA amounts disbursed to refugee claimants and to all Canadians are shown. Note that all refugee claimants are included here, even those who made their claim prior to 1999. Overall, between $10 billion to $13 billion were disbursed every year to SA recipients, based on the T5007 file.Note 16 Of that total, between $202 million and $338 million were disbursed to refugee claimants, accounting for 1.9% to 3.2% of the total, depending on the year.Note 17 This range of estimates is likely conservative, given the unlinked refugee claimants noted in the “Data and methodology” section (Section 2). Using the approach discussed in the sub-section “Defining social assistance and its incidence of receipt”Note 18 of this paper, the final two columns in Table 8 show the predicted value and predicted share of SA benefits paid to linked and unlinked refugee claimants. The predicted share of total aggregate SA benefits paid to refugee claimants rises to between 2.7% and 4.4% when this adjustment is made.
Thus, while SA rates are much higher among the refugee claimant population, this group represents a very small fraction of the Canadian population, and less than 5% of total SA payments are disbursed to their families.
How important is SA income for refugee claimants? To answer this question, the mean total and SA income generated at the family level are shown in Table 9. Since total income is required for this exercise, only refugee claimants who could be located on the T1FF are included in this analysis. Also, results are weighted by the number of family members, so that larger families account for a larger share of the results. For comparative purposes, results are also shown for all families.
Among refugee claimant families that received SA in year t+1, their average total family income typically ranged from about $19,000 to $22,000 (Table 9), while SA benefits generally ranged, on average, from $8,000 to $11,000. Hence, SA accounted for about 40% to 48% of total family income in year t+1. Average total family incomes were slightly higher in subsequent years, generally ranging from $21,000 to $27,000 by t+4 (again recalling that these figures pertain to the diminishing group of refugee claimants whose claims remained open). With average SA benefits at about $9,000 and average total family income increasing over time, the share of total family income consisting of SA benefits declined to between 33% and 42% by t+4.
It was shown earlier in Chart 2 that the incidence of SA receipt was more frequent among refugee claimants who arrived in the latter portion of the 2000s compared to those who arrived in the middle of the decade, possibly because of the recession that started in late 2008. The results in Table 9 suggest a similar trend in average SA amounts received, conditional on receiving SA.
SA eligibility is based on the need for income support. Hence, it is not surprising that the average total family income of refugee claimant families receiving SA was considerably lower than those of refugee claimant families not receiving SA. This difference was about $6,000 to $10,000 in year t+1, and about $10,000 to $15,000 in years t+3 and t+4.
Average SA benefits received by recipient refugee claimant families varied somewhat across socio-demographic characteristics (Table 10). As one would expect, average SA benefits were lowest in year t+1 among single individuals and highest among couples with more than two children, and lowest among young families.
Perhaps consistent with expectations, average SA benefits were somewhat higher among refugee claimants residing in families in which no one held a work permit than among those residing in families in which a work permit was held. This difference was generally in the range of $1,000 to $3,000. Across provinces, average SA benefits were generally highest in Ontario.
Finally, there was considerable variation in average SA benefits across countries of citizenship. Among the 2009 claimant cohort, average benefits were highest among refugee claimants from the Slovak Republic, Hungary and Somalia, and lowest among those from Cuba, Haiti, and India.
All measurements in this paper have been conducted at the family level. This is sensible since SA eligibility is determined at the family level. However, there may be some interest in detailing the individual characteristics of refugee claimants, and how these are correlated with SA receipt. For the most part, however, the results are largely in line with the family level analysis.
Table 11 shows the individual characteristics of refugee claimants in year t+1. Refugee claimants are somewhat more likely to be male than female (between 54% and 61% are male). About one-fifth to one-quarter of refugee claimants are under 19 years of age, while about 5% are aged 55 or older. Work permits were granted to roughly between 70% and 80% of refugee claimants over the age of 19.
Concerning the receipt of SA (Table 12), female refugee claimants were slightly more likely to do so than men—with a difference of about 3 to 5 percentage points. This is consistent with the findings from Tables 4 and 5 suggesting that single female and lone mothers were slightly more likely to receive SA than their male counterparts.
In most years, 80% to 90% of refugee claimants under age 19 were recipients of SA—either as principal claimants of SA or through familial ties, while this was generally the case for 75% to 80% of those aged 25 to 54.
Average social assistance benefit per refugee claimant
Finally, the average SA benefit paid to all individuals in the refugee claimant cohorts from 2005 to 2010, including non-beneficiaries of SA, was approximately $4,200 in t+1. The average SA benefit per individual in t+1 was somewhat lower among those in the 2005, 2006 and 2007 cohorts, at approximately $3,900, and somewhat higher among those in the 2008, 2009 and 2010 cohorts, at approximately $4,400 to $4,600.
Refugee claimants are an important part of the non-permanent resident population in Canada. They are also, by and large, a group characterized by vulnerability in terms of their recent circumstances and experiences, their weak ties to the Canadian workforce, and other factors. Financially, this vulnerability is evident in the relatively high rates of SA use. This report has drawn on newly linked administrative data files to paint a first picture of SA receipt among this population.
The study has been successful in linking approximately three-quarters of all refugee claimants to administrative files containing information on the annual receipt of SA. Of these linked claimants, between 65% and 85% rely on SA shortly after opening their refugee claim. Among those whose claims were still open after four years, some 25% to 40% were collecting SA. This figure is well above the Canadian average of about 8%.
The variation in SA rates tends to be strongest across two dimensions: province of residence and country of citizenship. Some differences exist across family type, age of oldest family member, and presence of a work permit within the family, but they are not nearly as pronounced. Between 85% and 90% of refugee claimants under the age of 19 are in a family that receives SA shortly after making a refugee claim.
Overall, between $10 billion and $13 billion were disbursed every year to all SA recipients in Canada. Given their relatively small size as a group, the dollar amount of SA paid to refugee claimant families amounted to between 1.9% and 4.4% of that total, depending on the year and on the treatment of unlinked cases.
Further research could examine the duration of SA spells among refugee claimants. In particular, which types of refugee claimant families tend to remain on SA for longer time periods? From a data-quality point of view, it would also be useful to be able to identify abandoned refugee claims (i.e., the claims of refugee claimants who have left the country, but whose cases have not yet been closed). In principle, the individuals involved in such cases are out of scope, but current data limitations preclude the possibility of dropping these cases from the analysis.