Text-only Preview

Giuseppe Burtini
March 20, 2011

Ethnicity, Immigration and the Income Gap in Canada
There are many proposed explanations of inequality in both economic literature and
popular media. We aim to test whether status as an immigrant in Canada alone, inde-
pendent of all the highly correlated variables, can be considered a determinant of income
inequality. Identifying determinants of income inequality is an important (and interest-
ing) problem, as inequality is recognized as a serious (and popular) social issue in most
rst world countries.
In Canada, from 1990 to 2009 immigration statistics show an average of over 200,000
permanent residents in the country and over 800,000 temporary residents. The over-
whelming majority of these immigrants have been from the Asia and Pacic region1.
Our hypothesis is that income inequality within Canada is determined by some function
of work experience (and quality of work experience), credentiality (especially recogniz-
ability) and education (and quality thereof), language barriers and cultural barriers and
a variable for race to capture eects related to discrimination. A previous paper by
Hum and Simpson (1999) found that in Canada there was no evidence of simple racism
for native-born visible minorities, suggesting that we should not expect signicant (eco-
nomically or statistically) results on the race variable. Our conjecture, which is largely
substantiated by the literature (Aydemir and Skuterud 2004 and Chiswick 1978), is that
language, legislative and cultural barriers are the main hindrance to immigrant employees
and that the wage dierential in that sense may be economically sound.
We will use publically available census data from the 2001 Canadian Census Micro-
data, which contains high quality data for variables including language spoken at home
and work, immigrant status, place of birth, generation in Canada, age of immigration,
occupation, total income. This will provide us just over 400 points of data for immigrants
(both permanent and temporary) and another 1600 for comparisons to non-immigrants
1A trend that may be slowing down over recent years as immigration from Europe and the United
States increases.

in the same population. The most important explanatory variable we expect is language
prociency, followed closely by education (and recognization of those credentials)
To capture a language prociency variable, we will work according to the recommenda-
tions in the 2001 Census Dictionary and create a variable that rates language prociency
high if an ocial language of Canada is spoken at home, medium if an ocial language
of Canada is spoken at work, but not at home, and low if neither2. We will also include
both year of (tenure in Canada) and age at immigration, the primary justication being
that immigrants who immigrated at a younger age may have acculturated more easily.
The years of education variable will be included in polynomial form, as we believe there
may be negative returns to excessive education3. We will take the education analysis
one step further and, under the assumption education occurs prior to work experience,
we assume that age at immigration combined with number of years of education can
be transformed to identify how much education was completed inside and outside of
After handling education and language prociency, we would also like to include vari-
ables for two other potential explanations: occupational regulation and immigrant en-
clave benets. For occupational regulation, we will classify the major eld of study
variable in to two broad categories, regulated and unregulated professions. We will also
provide summary information on the number of immigrants who are employed in pro-
fessions that are unrelated to their major eld of education, in particular comparing to
non-immigrants and comparing across regulation status. To attempt to capture enclave
benets, we will experiment with including a variable for the fraction of similar origin
2This method may miss an eect with regard to the value of French in most areas of Western Canada,
and the value of English in northern Quebec. It would be possible to correct for this issue by weighting
fraction of English/French speakers in the region, we however, will not do this as the data is only
available in metropolitan areas.
3We will similarly experiment with a interaction eect between location of education and amount, under
the hypothesis that immigrants who arrive in Canada highly educated may be subject to a stronger
negative acculturation eect.
4This methodology was used in a paper in Canadian Public Policy by Peter Li (Saskatchewan) to
discount immigrant educational experiences in terms of human capital.

immigrants in the subject's census metropolitan area.