Journal of Population Economics

, Volume 24, Issue 2, pp 451–475

Immigrant selection and short-term labor market outcomes by visa category

Authors

    • Faculty of Arts and Social SciencesSabanci University
Original Paper

DOI: 10.1007/s00148-009-0285-0

Cite this article as:
Aydemir, A. J Popul Econ (2011) 24: 451. doi:10.1007/s00148-009-0285-0

Abstract

This paper studies the efficacy of immigrant selection based on skill requirements in the Canadian context. The point system results in a much higher skill level than would otherwise be achieved by family preferences. This positive selection is achieved by directly selecting higher-skilled principal applicants who are assessed by the point system and also indirectly through higher-skilled spouses. However, due to difficulties in transfer of foreign human capital, immigrants admitted for their skills do not necessarily perform better in the labor market and important factors used to assess admissibility have very limited power to predict short-term labor market success.

Keywords

ImmigrationPoint systemVisa category

JEL Classification

J61J68

1 Introduction

Immigrant-receiving countries grant visas for permanent residence based on family ties, skill requirements, or humanitarian grounds. The allocation of visas across these alternative categories varies considerably across countries.

Family reunification is the cornerstone of the immigration policy in the US with majority of immigrants being admitted based on family ties (64% of all immigrants that arrived in 2001) that is followed by employment preferences (17%).1 Among the stock of immigrants in Germany in 2002, 42.8% arrived through family reunion and 16.3% under skill requirements, while in Denmark in 2001, 48.2% arrived through kinship and 5.7% arrived under skill requirements (Constant and Zimmermann 2005b). Other immigrant-receiving countries such as Canada and Australia, on the other hand, admit a much larger fraction of their immigrants based on skill requirements. During the 2000–2001 period in Canada, about 66% of immigrants were admitted under skill requirements and 27% under family class. During the same period in Australia, 51% of immigrants were admitted based on skill requirements and 36% under family ties.2

In immigrant-receiving countries, there is much debate about what fraction of immigrants should be admitted under each category and which factors should determine eligibility. Although specific rules change considerably over time, both in Canada and Australia, eligibility of immigrants based on skill requirements has been determined by a number of individual characteristics including age, education, experience, and language ability among others. In Canada, this selection mechanism, called the point system, was introduced during late 1960s for selecting immigrants under skilled-worker and business class categories.3,4 Canadian immigration laws also permit permanent residents or Canadian citizens to sponsor their family members (spouses, common-law partners, dependent or adopted children, parents, and grandparents) as immigrants under family class as long as sponsors are 18 years of age or older, live in Canada, and meet the income requirements. Eligibility of these immigrants depends strictly on kinship ties independent of any skill requirements. Similar provisions exist in other immigrants receiving countries such as the US and Australia.

The rationale for the skill-based selection mechanisms is to admit immigrants that can adapt to the labor market relatively easily and also help meet perceived demands for certain skill sets in the economy. Two important issues arise for assessing the efficacy of such selection mechanisms. First, do these mechanisms generate a higher-skilled immigrant flow and, if so, through which channels these outcomes are achieved. Second, how do immigrants that are selected based on skill requirements fare in the labor market compared to other immigrants. Comparison between skill-based immigrants and those in the family class is the most interesting since both historically and in ongoing discussions of potential policy changes the trade off has been between these two visa categories.

Human capital characteristics play a dominant role in selection decisions of skill-based immigrants. Therefore, a higher observable skill level, usually measured by years of schooling, is expected among this group than would otherwise be achieved through mainly family preferences. In terms of unobserved characteristics, however, it is not clear which group will have more favorable characteristics since these are primarily determined by self-selection among immigrants.

The performance of immigrants admitted under different visa categories in the host country labor market depends on skill levels and transferability of these skills. Family class migrants have access to family networks in the host country which facilitates access to crucial information regarding destination country labor market. This may allow them to base their decisions for migration on more reliable information and also may initiate investments in human capital prior to arrival that are valued in host country labor market. These types of networks may also help overcome barriers in the labor market through job contacts or better knowledge of processes leading to recognition of credentials. Skilled workers have much less access to these networks both prior to and following migration and their higher observable skill levels means that they are more likely to face skill transferability problems. Therefore, although government authorities aim a higher level of employability among skill-based immigrants, a priori it is not clear which group will have better initial outcomes and how these initial differences may evolve over time.5

This study addresses differences in observable skills across visa types and how these differences affect economic performance in the short run. A new data set, the Longitudinal Survey of Immigrants to Canada (LSIC), that recently became available in Canada is used for this purpose. LSIC is a unique data set with information on visa categories, a rich set of individual characteristics, and short-term labor market outcomes. Using this novel data set, this paper has three main contributions to the literature. First, it is shown that the point system generates a much higher-skilled immigrant flow than those admitted under family preferences. The main mechanism that generates this outcome is the selection of higher-skilled individuals within countries of origin rather than a shift in country of origin distribution toward those with higher mean skill levels. Second, in addition to directly changing the skill distribution by selecting higher-skilled principal applicants (PAs), the point system has an indirect effect on spouses’ skill distribution resulting in more educated spouses due to positive assortative matching. Third, favorable selection in human capital characteristics do not always translate into better labor market outcomes. This is because many of these characteristics, such as schooling and experience that are almost all acquired in source countries, have little or no return in the host country labor market and, hence, have very limited power in predicting short-term labor market outcomes.

The remainder of the paper is organized as follows. Following a review of the previous literature and discussion of data in Sections 2 and 3, Section 4 of the paper addresses selection across visa classes. Section 5 analyzes labor market outcomes, measured at 6 months and 2 years after arrival, focusing on labor force participation (LFP), employment, and earnings. The results stress that conclusions regarding relative performance of immigrants across visa classes differ across these margins. Finally, Section 6 summarizes the main findings of the analysis.

2 Previous research

There are several studies providing evidence on the relative success of immigrants admitted under different visa categories. Jasso and Rosenzweig (1995) find that, based on their occupations, skilled immigrants appear more favorably selected than immigrants entering as spouses of US citizens. For Australia, Chiswick and Miller (1992) and Wooden (1990) conclude that refugees have more difficulties in finding employment than other immigrants. These studies also report that the gap between groups narrows over time. Cobb-Clark (2000) finds that immigrants selected for their skills have higher LFP and employment rates 6 months after arrival in Australia. The differences in participation rates persist while that in employment rates dissipates by 18 months after arrival. Constant and Zimmermann (2005a) find that, in Germany, former refugees and those that arrive through family reunification are less likely to work full time compared to those who came through the employment channel. In the Danish context, however, they find that the legal status at entry does not play any significant role. In a companion paper focusing on earnings, Constant and Zimmermann (2005b) find that arriving through family reunion or as asylum-seekers or refugees has negative effects in both Germany and Denmark and legal status at entry have long-lasting effects.

The Canadian point system also received a lot of attention in discussions of immigration policy, but there is little direct evidence regarding its impact on immigrant characteristics and outcomes. Duleep and Regets (1992) and Borjas (1993) compare US and Canadian immigrants using census data and refer to the point system as a potential source of differences in immigrant characteristics between these countries. Both studies stress that the educational attainment of immigrants in Canada and the US from same source countries are very similar to each other. Borjas (1993) notes that since the late 1960s, following the introduction of the point system, Canada attracted a more educated immigrant flow relative to the US. The paper concludes that the point system alters the national-origin mix of the immigrant flow toward countries with higher average skills rather than attracting more skilled workers from a particular source country. Census data used by the above studies for US–Canada comparisons have no information on visa category and, therefore, only allows an assessment of the differences across host countries in average skill levels. Across country variation in skill levels is a function of both the attractiveness of the host countries for potential immigrants (i.e., who applies) and the selection processes (i.e., who is admitted; Aydemir 2006).6 Therefore, in order to attribute skill differentials across countries to immigration policy differences, one needs to assume that these countries would attract immigrants with similar skill levels under the same set of immigration rules. As opposed to the previous studies comparing immigrant characteristics across host countries, this study focuses on a single host country and analyzes variation in immigrant skills across visa types.

The only evidence on the impact of the Canadian point system on immigrant outcomes is by De Silva (1997) that examines earnings of skilled immigrants in Canada compared to assisted relatives and refugees. The author, using the Immigration Data Base (IMDB), finds that latter groups have lower annual earnings, but over time this gap gets smaller. Importantly, the study also finds that controlling for immigrant characteristics accounts for only a small portion of the earnings differential among the various classes.

3 Data

The LSIC is a survey of immigrants aged 15 years and older, who applied through a Canadian mission abroad, landed from abroad, and arrived in Canada between October 2000 and September 2001.7 This study uses wave 2 of the LSIC that surveyed around 9,000 immigrants approximately 2 years after arrival. LSIC wave 2 also provides information from a previous interview (wave 1) that took place approximately 6 months after arrival, providing a longitudinal feature. In the survey, one person per immigrant family was interviewed, but the survey also asked questions about the spouse of the person interviewed.

LSIC data contains rich information on education, training, labor market experience, language, and most importantly, the visa category of immigrants. Five visa categories can be identified: family class, skilled-worker class, business class, refugee class, and provincial nominees. Skilled workers are admitted under skill requirements and go through a points test. Business class immigrants are required to make investments in Canada. In addition to the financial requirements, they are subject to a relaxed points test. Family class and refugee class immigrants are not subject to the points test and are admitted based on family ties and humanitarian grounds, respectively. The number of observations for the last group, provincial nominees, is too small to analyze separately and this last group (0.7% of all immigrants) is excluded from the analysis.

Table 1 presents the visa category distribution of immigrants in the LSIC sample. About 61% of immigrants in the sample, age 15 and over, were skilled-worker class followed by 27% family class, 6% business class, and 7% refugees. When the sample is restricted to adult working age population, those 25 to 65 years old by the time of second interview, the fraction of skilled workers increases to 69% while those in family class is about 20%. Combining business class and skilled-worker class, around three in four immigrants were admitted through visas subject to skill requirements. For Canada, this historically marks one of the highest fractions over the 1980–2004 period.
Table 1

Distribution of immigrants across visa categories

 

Age 15 +

Age 25–65

Both sexes

Both sexes

Male

Female

Family class

27.0

20.4

14.9

26.1

Skilled-worker class

60.6

69.0

74.8

63.1

Business class

5.7

5.0

4.3

5.7

Refugee class

6.7

5.6

6.0

5.2

For family class immigrants, regulations require a “close relative” such as a spouse, a child, or a grandchild to be living in Canada who is willing to sponsor. Among the family class in the data, aged 25 to 65, 62% were admitted as spouses or fiancé, 35% as parents or grandparents, and 3% as other family members of the sponsor.8 Among skilled workers, on the other hand, 37% indicated they had a relative (not necessarily a close relative) already living in Canada when they arrived. The above numbers show that, while all family class immigrants had a family network in Canada when they arrived in the country, skilled workers were much less likely to have access to a similar network.9

When a family applies for migration, the family designates one of the individuals as the principal applicant and the remaining family members are called dependants. LSIC data provides the principal applicant information. In the case of skilled workers, principal applicant refers to the person who is assessed by the points test based on individual characteristics and the decision for the dependants rests on this assessment. Human capital characteristics of the skilled-worker dependants are not assessed by the points test. In the LSIC sample of 25- to 65-year-olds, the fraction of PAs among male skilled-worker and business class immigrants is around 0.87 and 0.90, respectively, while among females, the corresponding fractions are 0.33 and 0.15, respectively. Previous literature emphasized that, among skilled workers, only PAs are assessed and Cobb-Clark (2000), for example, restricts the analysis to the PAs. There may be, however, some association between skills of spouses due to positive assortative matching in the marital market. Therefore, this study includes both PAs and dependants to see if any of the selection patterns among PAs also carries over to the spouses.

The sample in the rest of the analysis is restricted to the adult working age population. The analysis is carried out separately by gender highlighting differences between males and females and by principal applicant status showing differences between PAs and their dependants.

4 Selection across visa categories

4.1 Education and language ability differentials

Educational attainment has been historically one of the main factors in the point system and the selection grid applicable to the cohort in this study allocated 16 points for education out of a maximum of 100 points. The pass mark was 70 points for skilled workers while it ranged between 25 and 40 points for business class (provided they satisfy other investment-related requirements). Given the lower pass mark, the impact of selection on resulting education levels is expected to be smaller among the business class compared to skilled workers.10

Table 2 panel A presents the educational distribution across visa categories using separate questions in the survey for the highest educational attainment and years of schooling. The results, based on the 6-month interview, are for the pooled sample of PAs and their dependents where dependants are assigned to the visa class of the principal applicant. Average years of schooling presented in the last row of the table shows that skilled workers have the highest schooling among both males and females followed by business class immigrants. Compared to family class immigrants, the schooling differential for skilled workers is 3.9 years among males and 3.4 years among females. Although most females are not PAs, this large education differential is noteworthy. Among males, interestingly, refugees have slightly higher education than the family class, although the difference is not statistically significant, while among females refugees are the least-educated group. In terms of the highest educational attainment, 87% of male skilled workers has a Bachelor’s degree or above, while 50% of business class, 31% of family class, and 24% of refugees have this level of education. In other words, a male skilled-worker class immigrant is almost three times more likely to hold a university degree than a family class immigrant. Similar differentials are also observed among females. These results clearly show that the level of education among immigrants admitted through skill requirements is much higher than other immigrants.
Table 2

Educational and language ability by visa category and gender

 

Male

Female

Family class

Skilled worker

Business class

Refugees

Family class

Skilled worker

Business class

Refugees

Panel A—education

  HS dropout

0.25

0.01*

0.04*

0.16*

0.28

0.02*

0.15*

0.33

  HS or trade school grad

0.26

0.03*

0.24

0.32

0.19

0.07*

0.29*

0.34*

  Some postsecondary education

0.19

0.10*

0.21

0.28*

0.19

0.20

0.24

0.19

  Bachelor’s degree

0.21

0.53*

0.37*

0.19

0.24

0.47*

0.28

0.12*

  Postgraduate or professional degree

0.10

0.34*

0.13

0.05*

0.10

0.24*

0.04*

0.02*

  Total

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

  Years of schooling

12.8

16.7*

14.9*

13.2

12.3

15.7*

13.4*

11.2*

Panel B—fraction with high language ability

  English speaking

0.47

0.74*

0.51

0.40

0.41

0.61*

0.29*

0.18*

  English writing

0.57

0.79*

0.58

0.49*

0.49

0.67*

0.38*

0.24*

  English reading

0.61

0.87*

0.65

0.54

0.51

0.76*

0.45

0.27*

  French speaking

0.07

0.16*

0.05

0.10

0.06

0.12*

0.03

0.06

  French writing

0.07

0.16*

0.04

0.09

0.08

0.13*

0.03*

0.09

  French reading

0.09

0.18*

0.06

0.11

0.09

0.16*

0.04*

0.10

Panel A presents fraction by highest level of educational attainment and mean years of schooling. Panel B presents fraction of immigrants with “high” language ability where “high” language ability refers to responses that indicate the language ability being well, very well, or mother tongue

*p < 0.05, significant differences for a given row in the table within each gender group relative to family class

Language ability is another factor important in the immigrant integration context and also one of the factors assessed by the point system. Points are awarded for each of the speaking, writing, and reading abilities in the official languages of English and French, accounting for a total of 15 points in the selection grid. As opposed to most data sets that only asks general ability in language, LSIC asked detailed questions about English and French language ability in speaking, writing, and reading. This information is unique not only because it relates to the selection grid but also because certain dimensions of language ability may be more important in the labor market integration of immigrants than others. For example, speaking ability may be more important in getting jobs if employers screen applicants on this dimension either because, on average, jobs are more likely to be intensive in oral communication or if poor speaking ability is regarded as a negative signal about the worker ability. Table 2 panel B presents the fraction of immigrants with “high” language ability, i.e., those who report their language skills being well, very well, or report the language to be their mother tongue.11 Male skilled workers report higher language ability in both official languages in all three dimensions relative to the family class immigrants. Male business class, family class immigrants, and refugees on the other hand have similar language abilities. Among females, skilled-worker class immigrants, similar to their male counterparts, report higher language ability than family class immigrants. Female business and refugee class immigrants, on the other hand, have lower language abilities. These results show that skilled-worker class immigrants not only have higher educational attainment but also report having higher language abilities.

4.2 Within-country and across-country components of skill differentials

The above differentials may be due to differences in national-origin composition of immigrants across visa categories or within source countries the point system may be selecting higher-skilled immigrants. Understanding which of these factors is primarily driving the differentials is important for policy as shifts in country of origin distribution may have different implications on average skills of immigrants under each scenario. Given the subjective nature of language ability question and space limitations, the paper focuses on schooling outcomes in this section to address this issue.

Small sample size does not allow an analysis by country of origin, therefore, individuals are aggregated into national-origin groups and the analysis is restricted to those in either family or skilled-worker classes. For each national-origin group j, let the mean education level be Sij in visa class i. The mean education level for visa class i is then a weighted average of Sij over N national-origin groups such that \(S_i =\sum\limits_{j=1}^N {p_{ij} S_{ij} } \) where pij is the fraction of immigrants from national-origin group j in visa class i. Let Ss and Sf be the average years of schooling among skilled-workers and family class immigrants, respectively. Then, the difference in years of schooling between these two classes can be decomposed into two components using the following form:
$$ \label{eq1} S_s -S_f =\left( {X_s -X_f } \right)\beta _f +X_s \left( {\beta _s -\beta _f } \right) $$
(1)
where Xs and Xf reflect the source country distribution in each visa class. The N × 1 vector of coefficients βs is the vector of average schooling levels across national-origin groups among skilled workers (where row βij is equivalent to Sij as defined above); βf is the corresponding vector for family class immigrants. The first term on the right-hand side of Eq. 1 refers to the schooling differential that is due to the differences in national-origin composition of immigrants. This term is called “across differential” in Table 3 where results from this Oaxaca decomposition of the schooling differentials are presented. The second term on the right-hand side, called “within differential,” is the differential between family and skilled-worker classes that is due to a different selection of immigrants within national-origin groups. Within national-origin group j, a more favorable selection of skilled workers relative to family class occurs when βsj > βfj.
Table 3

Decomposition of education differential between skilled-worker and family classes into within and across region of origin components

 

Panel A—all immigrants

Panel B—PAs

Panel C—dependants

Region of origin groups

Country of origin groups

Region of origin groups

Region of origin groups

Male

Female

Total

Total

Male

Female

Male

Female

Raw differential relative to family class (R)

3.98

3.49

3.85

4.42

4.16

3.18

2.27

4.69

Within (W; due to difference in mean education for a given origin)

3.69

3.15

3.50

4.20

3.81

3.14

2.08

4.10

Across (A; due to difference in origin composition)

0.29

0.34

0.35

0.22

0.35

0.04

0.18

0.59

Percent within (W/R)

92.7

90.3

91.0

95.0

91.6

98.8

91.9

87.5

Percent across (A/R)

7.3

9.7

9.0

5.0

8.4

1.2

8.1

12.5

The decomposition is carried out separately for males and females using 11 national-origin groups.12 The results in Table 3 panel A show that, for both males and females, over 90% of the schooling differential between skilled-worker and family class immigrants is due to differences within national-origin groups and less than 10% is due to differences in national-origin composition across the two classes. Within national-origin groups, country of origin compositions may be different across classes. Therefore, the same decomposition is carried out using a subset of countries of origin with large sample sizes and the results are presented in the last column of panel A, providing similar conclusions.13 Panels B and C also carry out the same decomposition separately for PAs and dependents and the results are very similar to those in panel A.

These results provide strong evidence that the point system generates a higher-skilled immigrant flow primarily by selecting more skilled immigrants within countries of origin rather than changing the country of origin composition.14 This has important implications for understanding the role of the point system and the economic opportunities in host countries in generating across-country variation in immigrant skills. For example, while among the immigrant stock in 2000–2001 average schooling levels within national-origin groups were, in general, higher in Canada relative to the US by about 1 year, for some source regions such as Asia and Europe immigrants in the US had higher skill levels (Aydemir and Sweetman 2007). This points out that the US could generate schooling levels among its immigrants close to or even higher than Canada without a point system and with its much larger emphasis on family reunification. This may be due to better economic opportunities in the US relative to Canada and underlines the importance of economic opportunities in host countries for generating high-skill immigrant flows in addition to designing policies for selecting high-skill immigrants.

4.3 Selection among principal applicants and spouses

The results in Table 2 show that skilled-worker males have 3.9 years more schooling than family class immigrants and, for females, although slightly lower, a remarkable 3.4 years differential exists. The similarity of the results for males and females is remarkable because, while close to 90% of skilled-worker males are PAs, hence are assessed by the point system, only about 30% of females are PAs yet they have very favorable characteristics as well. Selection among PAs and spouses is addressed in this section after restricting the sample to PAs and the spouses of PAs, leaving out a very small number of “other dependants.”15,16
Table 4

Educational distribution among PAs and spouses

 

Male

Female

Family class

Skilled worker

Business class

Refugees

Family class

Skilled worker

Business class

Refugees

Panel A—PAs

  HS dropout

0.25

0.01*

0.03*

0.17*

0.18

0.00*

0.09

0.34*

  HS or trade school grad

0.26

0.03*

0.26

0.32

0.20

0.02*

0.21

0.40*

  Some postsecondary education

0.18

0.09*

0.20

0.29*

0.22

0.11*

0.24

0.15

  Bachelor’s degree

0.21

0.54*

0.38*

0.18

0.28

0.51*

0.39

0.08*

  Postgraduate or professional degree

0.09

0.35*

0.13

0.05*

0.12

0.35*

0.07

0.03*

  Total

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

  Years of schooling

12.69

16.85*

14.90*

13.09

13.45

16.63*

13.16

11.04*

Panel B—spouses

  HS dropout

0.13

0.02*

0.03

0.28*

0.40

0.03*

0.12*

0.32*

  HS or trade school grad

0.23

0.07*

0.25

0.34

0.22

0.10*

0.34*

0.27

  Some postsecondary education

0.23

0.18*

0.28

0.13*

0.18

0.23*

0.22

0.24*

  Bachelor’s degree

0.29

0.49*

0.31

0.16*

0.17

0.45*

0.28*

0.15

  Postgraduate or professional degree

0.12

0.24*

0.13

0.10

0.04

0.19*

0.04

0.02*

  Years of schooling

13.68

15.95*

14.44

12.21*

10.75

15.57*

13.44*

11.59*

This table reports the years of schooling and the fraction by highest level of educational attainment within a visa category

*p < 0.05, significant differences within each gender group for a given level of education relative to family class

Table 4 presents the educational levels among PAs and spouses. First, this table shows that, in general, immigrants selected for their skills, skilled-workers or business class, have higher educational attainment than the family class among both PAs and spouses. For example, male skilled-worker PAs have over 4 years more schooling than male family class PAs. Interestingly, the differential between these two classes is larger among spouses than PAs with almost 5 years difference. Second, within the skilled-worker class, spouses have lower education than PAs for both males and females, which suggest that, within families, higher-educated spouses become PAs to increase the likelihood of passing the points test. This is not observed for other visa categories which is consistent with the fact that education has much less/no role in admission decisions.

The fact that not only PAs admitted under skill requirements but also their spouses have more favorable characteristics than the corresponding groups in the family class may be a result of positive assortative matching in the marital market leading to a positive correlation between skill levels of spouses. Table 5 investigates this correlation by presenting, for a given level of educational attainment of PAs, fraction of PAs married to a spouse with equal or higher educational attainment. The results show large variation in this fraction across classes with skilled-worker PAs having the highest fraction. For example, in panel A, among PAs with a BA degree, 67% of skilled workers are married to a spouse with the same or higher educational attainment, while among the family class, this is 53%, a difference of 14 percentage points.
Table 5

Fraction of PAs married to a spouse with equal or higher educational attainment, by visa category

Educational attainment of PA

Family class

Skilled-worker class

Business class

Refugees

Panel A—age 25–65

  HS dropout

1.00

1.00

1.00

1.00

  HS or trade school grad

0.71

0.86

0.80

0.74

  Some postsecondary education

0.70

0.77

0.71

0.55

  Bachelor’s degree

0.53

0.67

0.50

0.56

  Postgraduate or professional degree

0.27

0.37

0.25

0.14

Panel B—age 25–45

  HS dropout

1.00

1.00

1.00

1.00

  HS or trade school grad

0.86

0.87

0.87

0.74

  Some postsecondary education

0.76

0.78

0.66

0.57

  Bachelor’s degree

0.56

0.67

0.52

0.60

  Postgraduate or professional degree

0.27

0.38

0.26

0.16

Panel C—age 25–45, region of birth Asia

  HS dropout

1.00

1.00

  HS or trade school grad

0.84

0.96

  Some postsecondary education

0.71

0.85

  Bachelor’s degree

0.55

0.67

  Postgraduate or professional degree

0.17

0.33

In panel C, the sample sizes for business class and refugees are very small, therefore, no result is presented

There is some variation in mean age and region of origin composition across classes as shown in Table A.1 of the Appendix which may contribute to the education differences among spouses across visa classes. In order to make comparisons across more homogenous groups, panel B of Table 5 restricts the age to 25–45 that eliminates most of the difference in mean age between skilled-workers and family class and panel C further restricts the sample to those from Asia.17 Results from these last two panels provide the same conclusion that positive assortative matching is strongest among the skilled workers. Therefore, not only skilled-worker PAs are more educated, hence their spouses are likely to be more educated, but also conditional on the educational attainment of PAs, skilled-worker PAs have the most highly educated spouses. Hence, Tables 4 and 5 show that the point system has a direct effect that results in the selection of PAs with more favorable characteristics and also an indirect effect that creates a more favorable selection of spouses.

The results in panels B and C in Table 5 also show an interesting pattern where the positive sorting among skilled-worker class spouses relative to the family class gets much stronger as the PAs’ education level increases. In panel B, among high school graduates, skilled-worker PAs are only 1% more likely than family class PAs to have a spouse with equal or higher educational attainment. This probability increases to 2.6% among those with some postsecondary education, 19.6% among Bachelor’s degree holders, and 40.7% among those with postgraduate degrees.

The next section addresses to what extent these differences in observed characteristics across visa classes affect labor market outcomes in the short run.

5 Short-term labor market outcomes

This section analyzes the differences in labor market outcomes across visa categories relative to the family class immigrants. Three types of outcomes are discussed: LFP, employment, and earnings.

In the following analysis, the employment and weekly earnings refer to the reference week. However, the survey does not provide any reference week LFP information. For those not employed during the reference week, the survey either reports any previous employment for a time window defined for wave 1 (wave 2) as the time between landing (wave 1 interview) and wave 1 interview (wave 2 interview) or the main activity for those who never worked during this time window. In this study, the labor force participant is thus defined as someone who satisfies one of the following three conditions: (1) employed during the reference week, (2) not employed during reference week but reported previous employment, or (3) reported that main activity was “looking for work or establishing business.”18,19

5.1 Mean outcomes: labor force attachment and earnings

This section first presents average outcomes by visa class without any controls for either demographic or human capital characteristics. These results help answer the question whether large differences in observed human capital characteristics across visa classes documented in the previous sections are reflected in the labor market outcomes. The results also provide a benchmark for the latter analyses that explore the role of the observed human characteristics on the outcomes.

Pooling observations from waves 1 and 2, the following model is estimated:
$$ \label{eq2} Y_{it} =\alpha _0 +\alpha _1 X_i^1 +\alpha _2 X_{it}^2 +\alpha _3 X_{it}^3 +a_i +\varepsilon _{it} $$
(2)
where Yit is the labor market outcome for individual i at time period t, X1 is a set of dummy variables for visa classes, X2 is a set of dummies for year effects, and X3 is a set of variables that includes years since landing (YSM) and YSM interacted with visa class dummies.20 For LFP and employment outcomes, the functional form is a probit, while for the third outcome, it is a log-linear specification with log earnings as the dependent variable. Given the panel nature of the data, population-averaged models that allow for unobserved heterogeneity are estimated where ai is the individual specific unobserved effect.21 Estimation results are presented in Table 6 separately for PAs and their dependants by panels A and B, respectively. For LFP and employment outcomes, marginal effects are reported and robust standard errors are presented for all three outcomes in parenthesis.
Table 6

Assimilation profiles, marginal effects for LFP, and employment

 

LFP

Employment

Log earnings

Male

Female

Male

Female

Male

Female

Panel A—PAs

  Skilled worker

0.02 (0.02)

0.24* (0.04)

−0.08* (0.03)

−0.08 (0.05)

0.30* (0.04)

0.43* (0.09)

  Business class

−0.10 (0.06)

0.07 (0.13)

−0.57* (0.06)

−0.35 (0.19)

0.21 (0.20)

−0.38 (0.35)

  Refugee class

−0.33* (0.05)

−0.37* (0.07)

−0.04 (0.07)

0.03 (0.14)

−0.14* (0.07)

−0.23 (0.13)

  YSM

0.08* (0.03)

0.10 (0.06)

−0.04 (0.04)

0.06 (0.07)

0.15* (0.05)

0.33* (0.11)

  YSM × skilled worker

0.00 (0.02)

0.01 (0.03)

0.00 (0.03)

0.10* (0.04)

0.01 (0.03)

−0.06 (0.07)

  YSM × business class

0.01 (0.03)

0.06 (0.11)

0.25* (0.05)

0.23 (0.13)

0.05 (0.11)

0.21 (0.27)

  YSM × refugee class

0.09* (0.02)

0.11* (0.04)

−0.01 (0.05)

−0.02 (0.09)

−0.04 (0.04)

−0.01 (0.10)

Number

6,130

2,846

5,156

1,792

3,778

1,302

Panel B—dependants

  Skilled worker

0.19 (0.11)

0.09* (0.04)

−0.27* (0.08)

0.13 (0.08)

0.49* (0.18)

−0.08 (0.10)

  Business class

−0.20 (0.20)

−0.06 (0.06)

0.28* (0.05)

0.05 (0.11)

0.54 (0.34)

−0.17 (0.30)

  Refugee class

−0.41* (0.17)

−0.34* (0.07)

−0.02 (0.31)

0.01 (0.18)

−0.01 (0.15)

0.15 (0.20)

  YSM

0.16 (0.09)

0.04 (0.05)

−0.21 (0.14)

0.06 (0.09)

0.27 (0.18)

0.16 (0.14)

  YSM × skilled worker

−0.08 (0.06)

0.09* (0.03)

0.24* (0.10)

0.01 (0.06)

−0.18 (0.11)

0.20* (0.10)

  YSM × business class

0.07 (0.11)

0.06 (0.04)

−0.10 (0.11)

0.03 (0.08)

−0.16 (0.22)

0.17 (0.25)

  YSM × refugee class

0.08 (0.08)

0.22* (0.06)

0.01 (0.21)

0.10 (0.11)

−0.25 (0.23)

−0.23 (0.16)

Number

804

4,070

628

2,005

432

1,351

The reference group is family class immigrants and regressions include dummy variables for year effects but not other control variables. Results are from population-averaged models. Robust standard errors reported are reported in parentheses

*indicates significance at 5% level

Among males, LFP rates of skilled-worker PAs and dependants are not statistically different from their family class counterparts, employment rates are lower (8 and 27 percentage points, respectively) and earnings are substantially higher (about 30% and 49%, respectively). All visa classes register gains in participation rates and earnings levels as indicated by the estimated YSM coefficients. The YSM profiles for skilled workers are, in general, not statistically different from that of the family class with the exception of skilled-worker class dependants having a steeper profile. Among females, LFP rates are higher for both PAs and dependants compared to their counterparts in the family class (24 and 9 percentage points, respectively), employment rates are similar, and earnings are substantially higher for PAs but not for the dependants. Not all groups of females register gains in these outcomes over time, however, when they do, such as in the case of earnings for PAs, these gains are similar across visa types.

These results show that both male and female immigrants improved their labor market outcomes substantially over an 18-month period between the two interviews during which the Canadian resident population outcomes were stable as the last row of panels A to D of Electronic Supplementary Material Table A.2. Interestingly, much higher schooling levels for skilled workers relative to family class documented in the previous sections are not translated into more favorable outcomes especially for male PAs at the LFP and employment margins. It is, however, important to note the difference between male and female skilled workers that shows similar participation rates relative to family class for males but much higher ones for females. These differences between male and female skilled workers may reflect family labor supply decisions where males may be investing in human capital while females are taking jobs to support the family. This kind of “temporary” phenomenon is analyzed for immigrants in Australia by Cobb-Clark et al. (2005).

Results for LFP and employment outcomes also show that the groups that have the least favorable outcomes at entry register the largest gains. For example, refugees start out with the lowest participation rates but have the fastest growth in participation. Similarly, business class males start out with the lowest employment rate but have the highest growth rate which helps reduce the gaps over the 2-year period. This inverse relation, however, is not observed for weekly earnings, hence, the gaps persist 2 years after arrival.

5.2 Role of human capital characteristics

The previous sections documented large differences in human capital characteristics and labor market outcomes across visa classes. This section analyzes the role of demographic and human capital characteristics in explaining the variation in the labor market outcomes. An extension of the specification in Eq. 2 is estimated that pools data from waves 1 and 2:
$$ \label{eq3} Y_{it} =\alpha _0 +\alpha _1 X_i^1 +\alpha _2 X_{it}^2 +\alpha _3 X_{it}^3 +\alpha_4 X_{it}^4 +a_i +\varepsilon _{it} $$
(3)
where X1, X2, and X3 are identical to those in Eq. 2. X4 represents the additional controls that include years of schooling, experience and experience squared, controls for English and French speaking, writing, and reading ability, dummy variables for region of birth, region and metropolitan city of residence, marital status, province/city, and sex-specific unemployment rate at the time of interview.22 Visa class dummies in specification 3 capture differences in outcomes across classes after controlling for these characteristics where family class is the omitted category. Similar to the estimation of Eq. 2, probit models are estimated for LFP and employment outcomes, while for earnings, a log-linear model is estimated.23 The results of this estimation are presented in Table 7 for PAs and Table 8 for dependants. For binary outcomes, marginal effects are presented.
Table 7

PAs, assimilation profiles, and returns to human capital characteristics, marginal effects for LFP, and employment

 

LFP

Employment

Log earnings

Male

Female

Male

Female

Male

Female

Skilled worker

0.04 (0.03)

0.13* (0.04)

−0.06 (0.04)

−0.09 (0.06)

0.20* (0.05)

0.30* (0.09)

Business class

0.01 (0.04)

0.09 (0.12)

−0.50* (0.08)

−0.38 (0.22)

0.17 (0.19)

−0.20 (0.34)

Refugee class

−0.27* (0.06)

−0.34* (0.09)

−0.06 (0.07)

0.04 (0.14)

−0.12 (0.07)

−0.18 (0.15)

YSM

0.09* (0.03)

0.12* (0.06)

−0.04 (0.05)

0.07 (0.07)

0.15* (0.05)

0.35* (0.11)

YSM × skilled worker

−0.02 (0.02)

0.02 (0.03)

0.00 (0.03)

0.11* (0.04)

0.01 (0.03)

−0.04 (0.07)

YSM × business class

0.00 (0.03)

0.05 (0.12)

0.26* (0.05)

0.25 (0.14)

0.03 (0.11)

0.16 (0.27)

YSM × refugee class

0.06* (0.02)

0.13* (0.05)

−0.01 (0.05)

−0.02 (0.10)

−0.05 (0.05)

0.03 (0.11)

Years of schooling

−0.01* (0.00)

0.00 (0.00)

−0.01* (0.00)

0.00 (0.00)

0.01 (0.01)

0.02 (0.01)

Experience

0.00 (0.00)

0.00 (0.00)

−0.01* (0.00)

0.01 (0.00)

−0.01* (0.00)

−0.01 (0.01)

Experience squared/100

−0.01* (0.00)

−0.03* (0.01)

0.01 (0.01)

−0.02* (0.01)

0.01 (0.01)

0.00 (0.02)

English speaking

0.00 (0.01)

0.09* (0.03)

0.01 (0.01)

0.04 (0.03)

0.09* (0.02)

0.06 (0.05)

English reading

0.01 (0.01)

0.00 (0.03)

0.04* (0.02)

−0.07* (0.03)

0.01 (0.03)

−0.05 (0.06)

English writing

0.01 (0.01)

−0.07* (0.03)

−0.02 (0.02)

−0.01 (0.03)

0.03 (0.03)

0.08 (0.06)

French speaking

0.03 (0.02)

0.09 (0.06)

−0.01 (0.03)

0.02 (0.04)

−0.05 (0.05)

0.04 (0.11)

French reading

0.00 (0.02)

0.04 (0.05)

0.04 (0.03)

0.01 (0.04)

0.05 (0.04)

0.10 (0.10)

French writing

0.01 (0.02)

−0.01 (0.08)

−0.04 (0.03)

−0.04 (0.05)

−0.03 (0.06)

−0.10 (0.13)

Other controls

Yes

Yes

Yes

Yes

Yes

Yes

Number

6,126

2,846

5,152

1,792

3,775

1,302

Results are from population-averaged models. Each model includes dummy variables for region of birth, region and metropolitan city of residence, marital status, province/city, and sex-specific unemployment rate at the time of interview and at the time of arrival (monthly rates) and dummy variables for year effects. The reference group for visa category is family class immigrants. Robust standard errors are reported in parentheses

*indicates significance at 5% level

Table 8

Dependants, assimilation profiles, and returns to human capital characteristics, marginal effects for LFP, and employment

 

LFP

Employment

Log Earnings

Male

Female

Male

Female

Male

Female

Skilled worker

0.10 (0.13)

−0.11 (0.06)

−0.28* (0.07)

0.03 (0.10)

0.12 (0.25)

−0.21 (0.17)

Business class

−0.21 (0.20)

−0.15* (0.07)

0.18* (0.07)

−0.06 (0.14)

0.28 (0.33)

−0.16 (0.32)

Refugee class

−0.37 (0.28)

−0.39* (0.08)

0.01 (0.33)

−0.16 (0.24)

−0.36 (0.41)

0.29 (0.26)

YSM

0.17 (0.10)

0.07 (0.05)

−0.22 (0.15)

0.03 (0.09)

0.09 (0.19)

0.19 (0.15)

YSM × skilled worker

−0.09 (0.07)

0.08* (0.03)

0.24* (0.12)

0.02 (0.06)

−0.08 (0.13)

0.23* (0.11)

YSM × business class

0.09 (0.11)

0.04 (0.04)

−0.02 (0.15)

0.04 (0.09)

0.00 (0.20)

0.19 (0.24)

YSM × refugee class

0.07 (0.09)

0.21* (0.07)

−0.13 (0.22)

0.16 (0.12)

−0.09 (0.26)

−0.25 (0.17)

Years of schooling

0.00 (0.01)

−0.01* (0.00)

0.00 (0.01)

0.00 (0.00)

−0.01 (0.01)

−0.01 (0.01)

Experience

0.01 (0.01)

0.00 (0.00)

0.02* (0.01)

0.01* (0.00)

0.02 (0.02)

−0.02* (0.01)

Experience squared/100

−0.04 (0.02)

−0.02* (0.01)

−0.06* (0.02)

−0.03* (0.01)

−0.09 (0.05)

0.04 (0.03)

English speaking

0.01 (0.03)

0.06* (0.02)

−0.04 (0.04)

0.03 (0.02)

0.04 (0.06)

0.06 (0.05)

English reading

−0.05 (0.04)

−0.06* (0.03)

0.01 (0.05)

0.01 (0.03)

0.03 (0.10)

0.11* (0.05)

English writing

0.02 (0.03)

0.05* (0.03)

0.05 (0.05)

0.00 (0.02)

0.08 (0.09)

−0.09 (0.05)

French speaking

0.05 (0.05)

0.09* (0.04)

0.07 (0.10)

−0.06 (0.04)

0.04 (0.13)

0.03 (0.07)

French reading

−0.02 (0.05)

0.01 (0.04)

0.13 (0.07)

0.00 (0.05)

−0.12 (0.14)

0.17 (0.10)

French writing

−0.04 (0.06)

−0.01 (0.05)

−0.22 (0.13)

0.05 (0.05)

−0.23 (0.16)

−0.10 (0.10)

Other controls

Yes

Yes

Yes

Yes

Yes

Yes

Number

802

4,064

626

2,000

431

1,346

Results are from population-averaged models. Each model includes dummy variables for region of birth, region and metropolitan city of residence, marital status, province/city, and sex-specific unemployment rate at the time of interview and at the time of arrival (monthly rates) and dummy variables for year effects. The reference group for visa category is family class immigrants. Robust standard errors are reported in parentheses

*indicates significance at 5% level

A comparison of the estimated coefficients for visa types in Tables 7 and 8 to panels A and B of Table 6, respectively, shows that the introduction of these additional controls has limited explanatory power for explaining differences in entry outcomes across visa classes. For example, relative to the family class, 33 percentage points LFP disadvantage for male PA refugees in panel A of Table 6 drops to 27 percentage points in Table 7 with the addition of controls. Similarly, for the earnings outcome of the skilled-worker class immigrants, 30% advantage among male PAs relative to the family class is reduced to 15% and the 43% advantage for females is reduced to 35%. In other words, for PAs, the addition of controls explains about half of the earnings difference between skilled-worker and family class males and about one fifth of the difference for females. Also, after controlling for demographic and human capital characteristics, YSM profiles largely remain unchanged. These results are consistent with the findings of De Silva (1997) for Canada who reports that controlling for immigrant characteristics accounts for only a small portion of the earnings differential among various classes.

The returns to various characteristics are of special interest as they indicate how these factors, some of which assessed during the selection process, are related to future labor market success. Tables 7 and 8 report the effects of schooling, experience, and language variables. Experience pertains to almost all foreign experience, given that the immigrants in the sample have been in the country at most about 2 years. The returns to experience is either zero or, in some cases, negative. Similar results have been reported for recent immigrant cohorts to Canada by Aydemir and Skuterud (2005). Schooling has either zero or negative effect on participation and employment outcomes and a small positive effect on earnings.24 It is important to note that negative effects of schooling and experience on participation and employment are mostly observed for males but not for females, providing some support to possible family investment decisions where males invest in skill upgrading while females work. Both experience and schooling have large weights in selection decisions, yet over the short term neither seems to lead to any significant advantage in the labor market. This result should, however, be interpreted with caution. Immigrants with higher levels of schooling may be facing credentials recognition problems in the short term. Following possible investments in human capital and going through credentials recognition processes, the value of these characteristics may significantly rise over the long run.

Tables 7 and 8 also report returns to language skills. The point system evaluates speaking, writing, and reading abilities separately in both official languages, English and French. For PAs, coefficient estimates for speaking ability indicate a positive impact on labor market outcomes, while reading and writing abilities mostly have no significant effect.25 This suggests that speaking may be the most important dimension of language ability for LFP decisions, employment, and eventually for earnings. This stresses the importance of speaking ability for both immigrant selection and language training services.

The above results show that the three characteristics assessed by the point system, namely, education, experience, and language, do not have significant returns in the short run. Therefore, the contribution of these observable characteristics to explaining the differences in economic performance between visa classes is modest. Large unexplained differences that remain after controlling for demographic and human capital characteristics reflect the importance of unobserved differences in immigrant characteristics across visa classes. Factors such as quality of human capital characteristics, demand for them in the host country labor market, and access to networks may be playing crucial roles in labor market outcomes. Family class immigrants given their access to family networks, for example, may have advantages over skilled workers since they can use these networks for richer and more reliable information about the host country labor market while shaping the migration decisions and to overcome barriers in the labor market postmigration.

6 Conclusions

This paper studies the characteristics and short-term labor market outcomes of immigrants across visa categories in the Canadian context. There are three main findings. First, the paper documents that immigrants selected for their skills have much more favorable human capital characteristics. This is a result of the points system that generates a higher-skilled immigrant flow primarily by selecting more skilled immigrants within countries of origin rather than changing the country of origin composition toward countries with higher average skills.

Second, in addition to directly changing the skill distribution by choosing higher-skilled PAs, the points system also indirectly affects skills through higher-skilled spouses. The positive assortative matching among couples that leads to this result is found to be much stronger among skilled workers than other visa classes.

Third, while favorable selection in observed characteristics for skilled-worker PAs result in modest earnings advantages, it does not lead to higher participation and employment rates in the short term. This is because schooling and experience, almost all obtained abroad in this sample, have either zero or negative effect on LFP and employment and schooling has a small positive effect on weekly earnings. The fact that 2 years after arrival these characteristics do not significantly alter labor market outcomes indicates major difficulties in transfer of foreign human capital. However, the long-term impacts of both schooling and experience can potentially be different than the short-term ones if immigrants may be investing in human capital or going through credentials recognition processes. Also, after controlling for detailed demographic and human capital characteristics, large unexplained differences in labor market outcomes remain, indicating the important role of unobserved characteristics.

While the results in this paper underline that immigrant selection based on skill requirements similar to the point system may be very effective in changing the skill composition of immigrants, it is important to note that these types of screening mechanisms are usually limited to such observable characteristics as age, education, experience, and occupation. They are largely unable to affect selection in unobservables that are important in determining the quality and relevance of human capital immigrants bring to the host country. This may result in skill transferability problems or mismatches between the demand for specific skills in the host country and supply of them through immigration. The severity of these problems determines the extent to which host countries can benefit from immigrant selection mechanisms.

Footnotes
1

Legal Immigration, Fiscal Year 2001, Annual Report, US Department of Justice (August 2002).

 
2

Migration Program Statistics, Department of Immigration and Multicultural Affairs, Australian Government.

 
3

UK recently adopted a point system similar to that of Canada while there are ongoing discussions for introducing a similar system throughout the European Union and the US.

 
4

See the Electronic supplementary material for details of the Canadian point system.

 
5

Jasso and Rosenzweig (1995) also point out that, in the US context where employment-based immigrants are nominated by employers, employers may screen for short-term productivity while family members may screen for the long term. In the Canadian context, employer screening is much less relevant since skill-based immigrants can apply on their own without a requirement of a job offer.

 
6

Aydemir (2006) discusses the role of economic opportunities and the immigration policy on resulting immigrant characteristics. Using a sample of immigrants to Canada, a positive selection is found at the application stage among individuals from UK but a negative selection among those from the US.

 
7

“Landing” refers to the process of permanent residency taking effect. For an individual residing outside Canada, this occurs when the individual arrives in Canada through a port of entry. Generally, all applications for permanent residence must be made through Canadian missions abroad. Exceptional cases that are allowed to apply and become permanent residents while residing in Canada and are not required to leave and re-enter the country for landing are excluded from the LSIC sampling frame. These exceptions are discussed in more detail in Section 5.

 
8

Those applying under family class may be sponsored either by immigrants or native-born Canadians, but the data does not provide this information.

 
9

As long as the applicant is eligible for sponsorship under family class, there is no incentive to apply as a skilled worker which requires passing strict selection criteria. Those immigrated as skilled workers although they had relatives in Canada were most likely ineligible for family sponsorship.

 
10

Applications for immigration under any visa type include an on paper assessment of the information provided, as well as an in-person interview regarding the application. Upon request, applicants are required to submit original or certified documents regarding their education levels. Also, during the interview process, visa officers assess stated language proficiencies (Source: Applying for permanent residence in Canada: A self-assessment guide for independent applicants, Citizenship and Immigration Canada).

 
11

The remaining group consists of individuals who responded that their skills are fairly well, poor, or cannot speak the language. Language abilities refer to those reported at the 6-month interview.

 
12

These national-origin groups referring to broad regions of origin are dictated by the sample size. The 11 national-origin groups are North America, Central and South America, Caribbean and Bermuda, Western and Northern Europe, Eastern Europe, Southern Europe, Africa, West and Central Asia and Middle East, Eastern Asia, Southeast Asia and Oceania, and Southern Asia.

 
13

For this decomposition, countries of origin with a minimum of 100 observations are selected and seven of them satisfy this restriction. The mean number of observations per country is 515 observations, and the mean number of family class immigrants and skilled worker immigrants used to calculate Sij are 142 and 326, respectively. The subsample of seven countries makes up 60% of all family class immigrants and 56% of all skilled worker immigrants. These seven countries are UK, Iran, China, Philippines, India, Pakistan, and Sri Lanka. Male and female immigrants are pooled together for sample size considerations.

 
14

Note that Borjas (1993) focuses on the stock of immigrants who arrived starting from 1960s until 1981, while this paper focuses on a cohort that arrived over 2000–2001. Over time, both the importance of education in determining eligibility in the point system and the source country distribution of immigrants have changed considerably. These differences may explain different conclusions regarding the role of the point system.

 
15

Among the 25- to 65-year-olds, about 66.6% of the respondents are principal applicants, 32.7% are spouses of principal applicants, and about 0.7% are dependants other than a spouse.

 
16

The sample size for spouses of PAs is very small for some gender and visa category groups. In order to boost the sample size for this group to reduce sampling error and to provide a more comprehensive picture of education within families, the questions in the survey about spousal education levels are utilized. Therefore, information in panel B of Table 4 is obtained either from a survey respondent who is a spouse of a PA reporting his/her own education level or from a survey respondent who is a PA providing information about his/her spouse. The survey does not report whether the spouse is an immigrant or a Canadian-born, thus, especially for the family class, there is a possibility that the spouse is Canadian-born. When I use only the education information about respondents but not their dependants, the results are very similar for all visa categories but the family class whose education levels become significantly lower. All qualitative results, however, remain the same. Also note that, while survey respondents report both their highest educational attainment and years of schooling, they were only asked about the highest educational attainment of spouse but not the years of schooling. I assume that mean years of schooling for a given level of educational attainment is the same between survey respondents and their spouses.

 
17

Asia is the only source region with large enough sample sizes where reliable comparisons can be made between skilled workers and family class immigrants by educational attainment.

 
18

Individuals with missing labor force or employment status or those who decline reporting their wages are dropped from the sample. This results in a total of 394 observations being dropped out of 7,319 observations.

 
19

Table A.2 in the electronic supplementary material presents the mean labor market outcomes for each visa class using the LSIC data, as well as similarly defined outcomes for the resident population using monthly labor force surveys, adjusted to reflect the demographic characteristics and geographic distribution of immigrants. Resident population outcomes are presented mainly to provide a benchmark to assess broad patterns in the economy.

 
20

Using the LSIC survey, it is possible to calculate “time since landing” (landing referring to officially becoming an immigrant) which may be different than time since arrival as some individuals may already be living in the country before adjusting their status. These potential differences are important for assessing the relative performance of immigrants across classes since both the fraction that adjusts status within a visa class and time spent prior to adjusting the status may vary across visa categories. Available evidence suggests that there are very few skilled-workers or business class immigrants that adjust status (less than 2%) while almost half of refugees do so. Refugees adjusting their status are the ones that applied for refugee status within Canada and survey’s sampling design excludes these individuals. Therefore, the small number of remaining individuals that adjust their status among skilled workers and business class is unlikely to bias the overall results. As a further check, multivariate models were re-estimated by excluding from the sample individuals who ever lived in Canada before becoming permanent residents. Results following this exclusion are very similar to those presented in the paper.

 
21

Population-averaged models allow individuals’ error terms to be correlated over time. However, the heterogeneity term is assumed to be independent of individual characteristics. The estimated within-group correlation was high, suggesting the use of models that allows for heterogeneity. Fixed effects estimation is not possible in the probit context and random effects estimation could not produce reliable estimates for probit models as the results were very sensitive to the number of quadrature points. Random effects models for earnings outcomes are estimated with very similar results to those reported here from population averaged models.

 
22

The language controls are entered as continuous variables from 1 to 4, 1 being the lowest and 4 the highest language ability. Entering language variables as dummy variable sets results in very similar results. An alternative specification also replaced years of schooling variable with dummy variables indicating highest level of educational attainment and results were very similar to those presented here. Given the interest in this paper on the importance of factors assessed by the point system in predicting future outcomes, language ability and schooling as reported in the 6-month interview are used regardless of when the outcomes are measured. All other controls are as of the time of the interview corresponding to the outcome.

 
23

The earnings outcomes presented are based on OLS regressions that do not correct for selection into the labor force. This selectivity may be important particularly for females. To correct for sample selectivity in panel data context, I implemented the maximum likelihood estimator proposed by Zabel (1992); however, the estimation could not be carried out as the log likelihood in this case is not globally concave and iterations occasionally break down in this estimation as noted by Limdep software manual (pp. E23–E27). Alternatively, to get a sense of possible biases for earnings outcome, I estimated earnings regressions for each cross-section and produced parameter estimates without correction for selectivity and with correction for selectivity using a Heckit procedure. The conclusions from this cross-sectional analysis is that (1) returns to human capital characteristics, such as schooling and experience, are not sensitive to sample selection and (2) unexplained differences between classes still remain and, in some cases, become larger, after the correction for sample selection.

 
24

Very similar results for the impact of schooling are also found in specifications estimated using only a single wave of data.

 
25

Interacting French-speaking abilities with a dummy for the French-speaking province of Quebec does not change these conclusions.

 

Acknowledgements

I would like to thank the editor and two anonymous referees who provided valuable comments on earlier versions of the paper. I also benefited from comments received at presentations at Statistics Canada and CEA 2006 meetings. This paper represents the views of the author and does not necessarily reflect the opinion of any institution with which the author is affiliated.

Supplementary material

148_2009_285_MOESM1_ESM.pdf (250 kb)
(PDF 250 KB)

Copyright information

© Springer-Verlag 2009