Selectivity and Immigration in Canada: A Research Note

Derek Hum and Wayne Simpson

One important policy question respecting immigration is: What kind of immigrants should Canada accept? Should it be those with the best prospect of fast economic assimilation in the labour market? Our research sketches the earnings outcomes of immigration associated with the changing composition of immigrants’ admission classes. Specifically, we ask: what can we learn about the "selectivity issue" from examining the economic outcomes of immigrants to Canada from the Survey of Income and Dynamics (SLID), a survey conducted by Statistics Canada, and the Immigrant Data Base (IMDB), a file of administrative records pertaining to landings of immigrants?

There is much research suggesting that government policy determines the type of immigrants being admitted, which in turn, affects the "quality" of immigrants (Borjas, 1993; Reitz, 1998). For example, Reitz points out that Canada’s use of a point system makes its immigration more selective in terms of occupation than in terms of skill, whereas the U.S. attracts more highly educated immigrants in every origin group except Hispanics (Reitz, 1998). However, not all immigrants are subjected to the selection system, since Canada simultaneously tries to fulfill social and humanitarian objectives. It is clear that the composition of immigrants’ admission class in Canada changes considerably over time, with the 75 per cent in the independent or assisted relatives classes between 1967 and 1975 declining to less than 30 per cent by 1984 (Reitz, 1998). This percentage has fluctuated since then, rising again in the latter half of the 1990s.

Much has also been written on the earnings of immigrants, or earning disparities between the native-born population and immigrants entering Canada in different periods (Baker and Benjamin, 1994; Bloom, Grenier and Gunderson, 1995; de Silva, 1997; Grant, 1999). Recent research also suggests that immigrants have had an increasingly difficult time in the 1990s (Hum and Simpson 1999, 2001). The major aim of these studies was to investigate the possible reasons for divergence in earnings of immigrants and the Canadian-born, and the rate of convergence of this earnings gap. The performance of immigrants is therefore implicitly gauged against the "benchmark" of native-born Canadians. However, the literature has been relatively silent on the effect of the compositional structure of immigrants’ admission class on economic outcomes, although a recent assessment was made of Canada’s selection criteria in choosing skilled immigrants. (Citizenship and Immigration Canada 1998)

What can the IMDB and SLID data tell us?

Data on the compositional structure of immigrants can be obtained from the Immigration Data Base (IMDB), which provides a detailed breakdown of the admission categories of immigrants annually by linking immigration administrative records with income tax files. The IMDB covers the period 1980 to 1995.

The longitudinal performance of immigrants can be studied using the Survey of Labour and Income Dynamics (SLID). The SLID sample consists of more than 30,000 respondents and is a fairly comprehensive profile of immigrants and native-born Canadians.

The IMDB and SLID data bases collect valuable information on the immigrant economic experience in quite different ways. The larger IMDB data base covers about 69% of all immigrants between 20 and 64 who landed in Canada between 1980 and 1995. Immigrants must have filed at least one tax return since their arrival in Canada. The smaller SLID sample of households, on the other hand, also includes immigrants who do not file a tax return. Hence, these data bases provide two different, but complementary, views of the immigrant population. An obvious question, then, is how the data bases compare, and what can be learned from each.

Economic performance in IMDB and SLID

From the standpoint of the economy and integration in the labour market, important questions arise concerning the earnings and income performance of immigrants over time in comparison with their native counterparts. The IMDB and SLID data bases provide complementary views of this integration issue. Average earnings (and income) for males over the 1980-93 period are virtually identical, but the SLID estimates are again somewhat more volatile, consistent with the idea of sampling variability, but comparable coverage, in SLID for immigrant sub samples by year of migration.

For female immigrants who reported earnings, the average earnings and income over the 1980-93 period for IMDB and SLID are within 2% of each other, but the SLID series is also more volatile on an annual basis.

The IMDB allows us to look at the economic integration patterns by immigrant class. We grouped the 24 detailed immigrant classes into four major ones, based on the IMDB documentation: (1) family class, (2) economic principal applicant class, (3) economic spouse and dependent class, and (4) refugee class. For males, we find the following: (a) The economic principal applicant group is clearly a distinct group in the sense that it has higher earnings and somewhat faster earnings growth over time after arrival. (b) The refugee and family classes, on the other hand, are virtually indistinguishable. (c) The economic spouses and dependents class has similar earnings to the refugee and family classes at the time of entry into Canada, but their earnings growth flattens out and lags behind after about the first five years in Canada. Thus, there appear to be three distinct integration experiences for male immigrants based on category of admission.

For females, (d) the economic spouses and dependents, refugee and family classes are virtually indistinguishable.

An important consideration in interpreting these results, however, is that the apparent relationship between class of immigrant and earnings performance after entry may be due to other factors, such as better education and skills, language competence, or age, or some other dimension of human capital which enjoys higher market return. Exploratory investigation of aggregated data shows that: first, education matters (i.e., earnings are higher as education rises) and, second, the results are much the same for IMDB and SLID aside from the sampling variation in SLID.

Thus, an apparent partial correlation between class of immigrant and earnings may be attributable to other factors such as education, for instance. Unfortunately, neither the IMDB nor SLID public data base allows us to investigate this issue further, but for different reasons. The IMDB data base available to Metropolis Centres is not structured to allow researchers to explore the interrelationship between class, year of landing and factors such as, for example, education. It is impossible to determine whether Canada’s "selection" policies are effective, and if so, which components are pivotal. Access to more detailed IMDB data should allow this to be determined. This, clearly, is an item for future research.