Presentation to Canadian Employment Research Forum (CERF); Montreal, Concordia University, May 25 26. 2006. Revisiting Visible Minorities and Immigration Adjustment in Canada s Labour Markets. Derek Hum Wayne Simpson Department of Economics University of Manitoba Email: derek_hum@umanitoba.ca Email: simpson@ms.umanitoba.ca This research was partly funded by the Prairie Centre of Excellence for Research in Immigration and Integration (PCERII). We thank the Research Data Centre (Manitoba) of Statistics Canada for technical support. The views expressed are solely those of the authors.
Revisiting Visible Minorities and Immigration Adjustment in Canada s Labour Markets. 1. Introduction Economic opportunities vary widely among ethnic and racial groups in North America. The median family income of US blacks is only 60% that of whites. Hispanic families receive about 68% of that of whites, and Puerto Ricans as little as 45% (Leviatan et al, 1981: 238, 246). At the same time, Japanese and Chinese Americans have median incomes that are, respectively, 132% and 112% of the national average (Sowell, 1982: 46), illustrating that not all racial groups are necessarily disadvantaged relative to whites, or some notion of what Carlson and Schwartz (1988) have termed in reference to Canada a "charter cultural group". Canada attempts to alleviate disadvantage in labour markets for particular designated groups, deriving from its Employment Equity Act (Short Title, Assented to 15 th December, 1995), the purpose of which is: to achieve equality in the workplace so that no person shall be denied employment opportunities or benefits for reasons unrelated to ability and, to correct the conditions of disadvantage in employment experienced by women, aboriginal peoples, persons with disabilities and members of visible minorities. (Section 2). The Act clarifies the meaning of visible minorities to be:
persons, other than aboriginal peoples, who are non- Caucasian in race or non-white in colour;. Although employment equity in Canada includes aboriginal peoples and persons with disabilities, this paper concentrates principally on visible minorities, in conjunction with gender and immigration status. 1 Economic disparity among visible minorities is often attributed to discrimination. But how much of the disparity is actually the result of productivity-related considerations such as education, work experience, or language proficiency in Canada s two dominant commercial languages, French in Quebec, or English in other Canadian provinces? 2 Another important consideration for visible minority Canadians is their place of birth, and if foreign born, the length of time they have lived or worked in Canada. There is ample evidence that immigrants face economic disadvantage upon arrival, but this disadvantage declines as they assimilate in the workplace (Borjas, 1994). It is therefore necessary to account for immigration circumstances when considering visible minorities since two of every three new immigrants to Canada are visible minorities. Further, the visible minority population will grow more rapidly than total population to 2016 (Kalbach et.al, 1993: 8, 24 1 There are unique labour market issues associated with Aboriginals. Additionally, there are too few aboriginals in our sample to afford statistical analysis and the issue of immigrant aboriginals is not relevant. Persons with disabilities have been examined elsewhere by the authors (Hum and Simpson 1996). 2 US blacks speak English but still earn only 60-70% of that of whites so productivity factors beyond language are obviously important. Language is a perennial issue in Canada but is not our main concern in this paper. We do, however, allow for linguistic differences in our analysis
ff.). Information concerning visible minorities reported by media or advocacy groups is typically highly aggregated and conceals much detail. Consider, for example, Table 1 below, which displays annual earnings and hourly wage rates for both whites and various visible minority groups in Canada. 3 Visible minorities have annual earnings of $25,880 and an hourly wage rate of $17.89 compared to earnings of $29.992 and a wage rate of $19.43 for whites. In other words, visible minorities as a group suffer a nearly 9% wage disadvantage and a 16% earnings disadvantage. [Table 1 about here] While all visible minority groups in Table 1 have lower earnings, and all but the Chinese have lower wages than whites, there is clearly variation among the different groups. 4 One complication, immediately apparent, is the difference in the number of hours worked. For example, although non-chinese Orientals have lower wage rates, they work substantially more hours. However, Arabs work much fewer hours and of earnings. For an assessment of earnings differentials by linguistic groups in Quebec, see Shapiro and Stelcner (1997). 3 The data are from SLID, and described in a later section. 4 Our groupings also mask considerable variation. For example, Japanese have annual earnings well above the comparable figure for whites, although the small sample size makes comparison
earn much lower amounts ($21,244 for Arabs cf. $28,244 for Non- Chinese Orientals), despite having a higher wage rate than non-chinese Orientals. We believe wage rates are a better measure of labour market opportunity for paid workers than annual earnings (Christofides and Swidinsky, 1994:35), since earnings represent a combination of individual effort (hours worked) and offered rates of pay. Table 1 reveals little variation in average age, but more variation in the proportion of males in each visible minority group. Consequently, given that men are generally paid more than women, this would be reflected in the average hourly wage rate and earnings. More noteworthy, there is extremely wide variation in immigration status; only 10.6% of the white sample is immigrant while between 79% and 96.5% of visible minority groups comprise immigrants. If immigration status is associated with various labour market disadvantages, then these factors, visible minority membership aside, may be responsible for the observed wage differentials. In short, Table 1 is a salutary warning against hasty generalization about the structure of opportunities in the Canadian labour market based upon a simple comparison of whites and visible minorities. unreliable. Small sample sizes for Koreans, Japanese, Southeast Asians, Filipinos and Oceanic
This essay investigates wage gaps for different visible minorities in Canada and its evolution during the past decade, using data from the Survey of Labour and Income Dynamics (SLID). In addition to considering the usual determinants of wages such as age, education, residential location and the like, we also examine the issue of gender. Not only are women a designated group in the Employment Equity Act, it is important to query the double negative ; that is, whether visible minority women experience disadvantage in addition to that occasioned by gender. Given the significance of foreign born status noted above, we also explore immigrant status among visible minorities in Canada, also by gender as well. Our results have implications for both employment equity and immigration policy. 2. Previous Research in Canada Christofides and Swidinsky examine the 1989 Labour Market Activity Survey (LMAS), and report that minority women are especially disadvantaged. They use a dichomous (dummy) variable derived from a self-perception question to capture visible minority membership and acknowledge their data do not allow [them] to conduct an analysis members force us to group them together as non-chinese Orientals.
of individual minority groups" (Christophides and Swidinsky, 1994: 46). 5 Consequently, it is not possible to determine whether certain visible minority groups earn less than their white counterparts while other groups earn more. Furthermore, Christophides and Swidinsky (1994:39) conclude that immigrants are "generally not disadvantaged in the Canadian labour market". This finding appears at odds with results by Bloom, Grenier and Gunderson (1995), who apply a model developed by Chiswick (1978) and Borjas (1985) to pooled Canadian census data from 1971, 1981 and 1986 to examine immigrants. The Chiswick-Borjas model explains earnings as a function of human capital factors, such as education and potential experience (age less years of education), labour market effort (number of weeks worked and the number of hours worked per week), and immigration variables. The immigration variables include a dummy indicator distinguishing those born outside Canada to measure "the entry effect," and the number of years since migration to Canada to measure 5 A dummy variable simply indicates the presence or absence of a particular qualitative trait such as female gender, disability status, or in this case, visible minority membership. While useful for determining whether the particular characteristic trait is statistically important, the dummy variable does not capture any degree or extent of the characteristic trait. Thus, a dichotomous or dummy variable indicates whether or not a person is a visible minority but cannot distinguish among different visible minority groups such as Chinese or Arabs, etc.
"the assimilation effect." 6 Bloom et al find a negative entry effect (earnings for immigrants upon entry into Canada are less than native born Canadians) and a positive assimilation effect (earnings of immigrants tend to grow faster than average Canadian earnings). Their estimates imply that earnings of immigrants catch up to the native born in about 25 years (the assimilation effect). The authors comment, however, that the Canadian labour market has had difficulty in assimilating more recent cohorts of immigrants. Possible reasons suggested include: a reduced absorptive economy, a reduction in immigrant skills, and "increased discrimination as the composition of immigrants changed towards more visible minorities" (p.1000). But, as Bloom et al do not incorporate visible minority status in their analysis, it is not possible to confirm their suggestion of discrimination towards visible minorities. DeSilva (1996) uses Census data to examine the earnings of immigrants, many of whom are visible minorities. He concludes that earnings differences for visible minority immigrants can be explained by 6 The entry effect is the difference in wages between immigrants upon first landing (when years since migration is zero) and those native born. The assimilation effect is the annual rate of decline in this difference after entry. Bloom et al also use a series of dummy variables to estimate cohort effects for immigrants. In our study, which is confined to a single cross section, we can only use years since migration since it is perfectly collinear with cohort effects at any given point in time.
differences in the quality of seemingly-identical educational qualifications. His conclusion is based upon the fact that virtually no earnings differential (and hence discrimination) was found between Canadian-born visible minorities and Canadian-born non-visible minorities. Again DeSilva makes no distinction among different visible minority groups; therefore it remains possible that positive and negative earning differentials for different visible minority groups offset one another. Beach and Worswick (1993) employ the Job Mobility Survey for females aged 25-64 to determine if there exists a "double-negative" effect; that is, whether immigrant women suffer a disadvantage in addition to any disadvantage due to gender. Although they find no "across the board" effect, they do report a double-negative effect that is "quite marked for highly educated immigrant women" (p.35). Consequently, the gender dimension of employment opportunities in Canada cannot be ignored. Some authors have explored differences among visible minorities groups themselves. Pendakur and Pendakur (1998) employ the 1991 PUMF and report considerable differences in earnings between whites and visible minority groups, as well as among visible minority groups. Blacks males (native born and immigrants) face large earnings penalties.
However, Canadian-born Black women face no earnings penalty. Hum and Simpson (1999) employ the 1993 SLID for similar purposes, but focus upon wage offers. In sum, Canadian research exploring labour market disadvantage often focus on visible minority status, or gender, or immigrant adjustment. Immigrants to Canada are increasingly visible minorities, and a consensus is emerging that recent immigrants are not experiencing the same success as in the past. The reasons suggested are numerous, and the specific contribution of employment equity policies themselves remains largely unevaluated. We revisit some earlier results of Hum and Simpson (1999) concerning wages and visible minorities, and describe what changes have taken place during the past decade. 7 3. Data and Sample Description Not only does SLID provide more detailed information on labour market activity than Census information, the master file also allows a fine grain examination of visible minority groups. [Table 2 about here]
Specifically, our SLID sample consists of 26,759 respondents with reported earnings in 2002 and who were not a student or aboriginal. 8 For these respondents we have the following information: annual earnings and a composite wage rate for all jobs held in 2002, labour market activity (hours worked per week and weeks worked), human capital characteristics (full-time equivalent years of schooling completed, whether educated in Canada, and full-year full-time equivalent years of work experience), gender, mother tongue, location (region of residence and size of community), visible minority status by group, and immigration details (including years since immigration and age at immigration). The weighted sample means are reported in Table 2. Table 2 represents a fairly comprehensive profile of Canadians. Not only is there classification by visible minority group, there is also detail on immigration status, particularly the number of years spent in Canada since immigration. This is particularly important, since there is invariably a period of adjustment for immigrants and adjustment patterns differ across visible minority groups. In addition, information 7 See Hum and Simpson (2004a) for a recent survey of empirical evidence on the economic assimilation of immigrants in Canada.
on the age at immigration, and whether or not an immigrant was educated in Canada, may be important given earlier findings (DeSilva, 1996). The SLID also contains superior information on labour market activity. Rather than relying on age (minus years of schooling) as a proxy for potential work experience, SLID provides a direct measure of full-time full-year equivalent work experience that takes into account interrupted work careers. 9 Finally, SLID contains detailed demographic information and a composite wage determined from all jobs held over an entire year, not merely a single job during a reference week. These details enable us to explore visible minority status and wages in some depth. In particular, we can account for characteristics pertaining to accumulated human capital, current labour market activity, immigration, gender, language, and location that affect wage rates and, as a result, confound generalizations about the relationship between colour and wages glimpsed from simple comparisons such as those displayed in Table 1. 4. A Preliminary Look at Visible Minority Wages 8 Students are traditionally excluded from studies of earnings because their primary activity is education rather than work. Aboriginals are excluded because of incomplete coverage, particularly of the population on reserves.
We indicated that a simple tabulation of earnings for different visible minority groups in Canada ignores many factors, such as education, work experience, age, place of residence, that also determine wages and earnings. How would incorporation of these considerations change the portrait given by Table 1? Multiple regression analysis can be used to determine what wage gaps exist when factors in addition to visible minority status are taken into account. 10 We consider the effect for men and women combined, using, first, a simple dummy variable to indicate visible minority status and, alternatively, a series of dummy variables to represent different visible minority groups. 11 We find that simply being a visible minority results in wages about 6% less than other Canadians after allowing for the effects of accumulated human capital, current labour market activity, immigration, gender, language, and location, etc. This estimate is significantly different from zero. 12 For particular visible minority 9 Hum and Simpson (2004b) demonstrate that using potential experience rather than actual experience exaggerates both the disruption and recovery caused by immigration. 10 The logarithm of the wage rate provided a better fit than the wage rate level in preliminary results that are available from the authors upon request. 11 These basic results are not reported but available from the authors on request. Results of a even more detailed nature are given in an appendix. 12 We shall indicate different levels of significance in this study. The 5% level is adopted as conventional with the 10% level termed slightly significant and the 1% level highly significant. This may be confusing to some readers, but we believe it is necessary to adopt various levels to distinguish changes that occurred over time and results that differ due to sample sizes and varying conventions.
groups, we find the following (again, all statistically significant): Blacks receive 13% less than non visible minority Canadians, Indo-Pakistani persons receive 5.5% less, Latin Americans receive 12% less, and non- Chinese Orientals receive about 8% less. Chinese, however, do not receive less. These wage differences are quite different for some groups depicted in Table 1. For example, the "gross" differences in Table 1 for the Indo-Pakistani group (-21%) and the Latin American group (-23.1%) are much larger than the respective "net" differences of -5.6% and -12% estimated by multivariate analysis. This suggests that much of the gross difference (of Table 1) can be explained by factors other than visible minority membership. 13 5. A Closer Look at the Wage Differences The above results can be explored more closely. One obvious refinement is to examine men and women separately, since labour market experiences undeniably differ between men and women. Another enhancement is to compoare different visible minority groups to 13 We also corrected for any bias arising from considering only workers, as opposed to potential workers (those willing to work). After this adjustment, the estimated effects of visible minority status are much the same.
determine the magnitude of disadvantage for each group. Accordingly, we estimate separate wage regressions for men and women by visible minority status. In addition to better measures for selected variables, our results also differ from earlier research by including information on the circumstances of immigrants. This omission from earlier studies is important since immigrant status is found to be (highly) significant only for visible minority men. 14 The wages of immigrants who are not visible minorities converge most quickly, reaching equivalence with their native born Canadian counterparts for both women and men. For visible minorities however, convergence is slower. Therefore, while visible minority status affects the speed of convergence of immigrant wages to those of the native born, it is clearly immigration status that is important in determining wage differences between members of visible minorities and other Canadians. 15 We also estimate separate regressions for immigrants and for native-born Canadians while allowing for differences in visible minority membership. Our results are summarized in Table 3, which capture the most 14 We ignore here the well understood bias arising from heterogeneous immigrant cohorts and what Borjas (1985) terms declining cohort quality since we are only using immigrant status as a control variable to improve our estimates of wage differences by visible minority status. See Hum and Simpson (2004a) for a proper discussion of this bias and a survey of its implications for the analysis of immigrant labour market integration for Canada.
interesting and important results. (Full details are presented in an Appendix.) [Table 3 about here ] After accounting for other factors, including immigrant status, membership in a visible minority (in 2002) is significant principally for immigrant men. For Canadian born men, visible minority membership is generally insignificant (at the conventional 5% significance level), as it is for both immigrant and native-born women. Among visible minority groups, there is a significant wage disadvantage for black men of approximately 22%, whether immigrants or native born. There are also significant disadvantages for immigrant men who are members of the Indo-Pakistani group (13.1%), and the non-chinese Oriental group (14.0%). Adopting a less strict level of significance (10% rather than 5%) reveals disadvantage for Arab men (13.7%) and Latin American male immigrants (17.7%). But note that native born Latin American men experience disadvantage of 38.8% at the highest level of significance. In sum, Canadian born Latin American men currently suffer a 15 These results do not adjust for across cohort bias or bias associated with the different experiences of successive immigrant cohorts. See Hum and Simpson (2004) for a discussion of this issue.
disadvantage that is even larger than that of Blacks in Canada. Interestingly, native born Indo-Pakistani women enjoy a large wage premium of nearly 27%. All other groups are individually insignificant. The economic circumstances of Blacks in Canada deserve further comment, especially since the results appear so startling and robust. Past empirical studies of the Canadian labour market have not especially concentrated on Blacks per se, but have rather examined the earnings of all visible minority groups in the context of changing origins and racial composition of immigration to Canada since the 1970s, specifically the shift towards visible minority immigrants from non-english-speaking nations. Baker and Benjamin (1994) compared black men to native born men using the 1971, 1981 and 1986 Census Public Use Micro Data Files. They find significantly lower earnings for blacks as a group, combining both immigrants and native-born workers, but cannot distinguish a specific black immigrant effect in a pooled sample of the foreign and native born. A subsequent study by Pendakur and Pendakur (1998) using the 1991 PUMF data also found significant earnings penalties for black men, whether Canadian born or immigrant. These results are consistent with Hum and Simpson (1999), who find significant lower earnings for visible minority men as a group relative to
native-born men and significantly lower earnings for native-born black men compared to white men in the Survey of Labour and Income Dynamics. That is, they find significantly lower earnings for both immigrant and native-born black men, although they do not test to see if these differences are distinctive. 16 The situation of Blacks (especially native born men) in Canada needs more high quality empirical research, especially since recent findings confirm that, even after accounting for immigrant status and gender, Canadian born black men face, by far, the largest (statistically significant) earnings disadvantage. 6. Revisiting Visible Minority Opportunities: Quo Vadis? A decade has passed since the 1995 Employment Equity Act became part of Canada s economic policy culture. Canada continues to be an immigrant-receiving country. How have things changed? Do 16 Duleep and Regets (1992) find from the 1981 Census that the entry earnings of Chinese immigrants are 53% below native earnings while the entry earnings of British immigrants are 13% above native earnings, suggesting substantial differences in cohort effects related to language and visible minority status. Schaafsma and Sweetman (2001) show that members of a visible minority who immigrate at a younger age have higher age-earnings profiles than their older counterparts, particularly immigrants after the age of 35, suggesting that acculturation plays an important role in determining immigrant economic success. Since the changing composition of migrants toward non-european immigrants conflates language, visible minority status, and a host of economic and cultural differences, distinguishing these different effects is a demanding, but potentially important, task for future research on visible minority earnings.
visible minorities fare better or worse today than a decade ago? The earlier examination of wage disadvantage of visible minorities by Hum and Simpson (1999) using the 1993 SLID Master file data (sample size = 11,344) can provide a short historical perspective. The 1993 data give a benchmark snapshot immediately prior to the 1995 Employment Equity Act, while this essay examines 2002 data. 17 That earlier study reported similar findings in the sense that visible minority immigrants faced economic disadvantages to various degrees. However, disadvantage for Arabs and Latin Americans was not statistically significant (at the conventional 5% level) in 1993. A conventional statistical significant wage penalty of 17.3% was found for Chinese immigrants (but not native born Chinese men) in 1993. (See Table 3, lower panel) In contrast, the 2002 data find slightly statistically significant penalties of 13.7% and 17.7%, respectively, for Arab and Latin American men, as well as a significant penalty of 38.8% for native born Latin American men (see Table 3, upper panel). These results, we believe, are primarily due to the larger sample size (of 26,759) available for the present study rather than the sudden emergence of penalties for 17 This discussion should not be read as an evaluation of the Employment Equity Act. Rather, we are content with a modest general description of the change in circumstances of visible minorities in the past decade. A full evaluation of the Employment Equity Act, or an analysis of
these two groups since 1993. To highlight this feature, we note that estimated wage gaps for Latin American and Arab male immigrants were statistically insignificant (at 5% level) in 1993. In 2002 however, these wage penalties are statistically significant at the 10% level. Furthermore, the wage penalty for native born Latin American men is not only statistically significant at the 1% level, but also large. In short, Latin American men, whether native born or immigrant, are now experiencing substantial wage penalties, though not to the degree faced by Black men in Canada. As previously noted, black men have a wage penalty of about 22%, whether immigrant and native born Chinese male immigrants no longer appear to have a wage penalty in 2002. This result cannot simply be due to larger sample size however, since the smaller sample available a decade earlier was sufficient to establish conventional statistical significance. Rather, we believe this change is due to aspects of selection in the intervening period with respect to Chinese immigrants. Chinese immigrants to Canada since 1993 appear to possess traits in terms of education, human capital skills, as well as financial and entrepreneurial capital means that contribute to their economic performance, and which are not fully how different visible minority groups have fared in the labour markets in the past decade is
captured by the explanatory variables included in our regressions. 18 As a result, the selective immigration by this visible minority group may have eliminated the previously conventional significant wage penalty. While this is unlikely to be the entire explanation, we suggest that this composition effect is a large part of it. The general view that wage penalties are more problematic for men than for women has not appreciably changed. However, inspection of Table 3 (both panels) reveals that Non-Chinese oriental women immigrants no longer have a wage penalty so the statement that immigrant women do not experience wage penalties now appear to be unambiguous supported for all visible minority groups. At the same time, while few visible minority native born women have a (statistically significant) advantage or penalty, it is fascinating to discover that native born Indo-Pakistani women now receive a wage premium of nearly 24%, (and highly significant at 1%). This warrants more detailed beyond the scope of the present essay. 18 We realize some may regard this as too sweeping a statement, but the flight of people and capital from Hong Kong prior to 1997 as China prepared to reclaim Hong Kong, as well as new investment from Chinese entrepreneurs encouraged by Canadian immigration policies, must certainly be factors on casual inspection. Detailed examination of the pattern of immigration to Canada by Chinese during this period is beyond the scope of the present essay. We merely note that there are candidate explanations for why Chinese male immigrants are no longer disadvantaged in the 2002 data while being so in the 1993 data. The proportion of the Chinese population in Canada that is immigrant increased from 72.5% (authors calculation) to 83.3% (Table 1). Li (1993) provides a discussion of investment and business in Canada by Chinese immigrants but his discussion ends just before the 1993 Employment Equity Act.
research on the labour market experiences of this particular group. If we were to summarize the changes over the last decade in terms of a scorecard of progress, then Chinese male immigrants have made positive gains, and Arab and Latin American men are now shown to be disadvantaged. As in the past decade, black males continue to experience substantial disadvantage, whether Canadian born or not. Indeed, there is no longer any distinguishable difference in the amount of wage disadvantage between immigrant and native born black men in 2002, while previously, native born black males fared slightly better than immigrant black males. 7. Summary and Policy Implications The extent to which visible minorities participate in the Canadian economy is an important policy issue. Together with women, aboriginal peoples and persons with disabilities, visible minorities are a designated disadvantaged group under federal employment equity legislation. While these four groups identified by employment equity legislation are very different, the matter of racial discrimination is central to visible minorities and strikes at the heart, perhaps, of Canada's self image as a tolerant and liberal democracy. At the same time, immigrants to Canada
increasingly come from non-white countries of origin; hence Canada's self image as an accommodating immigrant-welcoming nation is also at stake. Our research reveals the danger of merely collating information on visible minorities to draw inferences concerning discrimination by colour. This type of exercise, typified by Table 1, is incomplete --- and worse, misleading --- because it combines all visible minority individuals without distinguishing their colour, ethnic origin, education, work experience or degree of assimilation into the Canadian labour stream. A more accurate picture is Table 3, where it is apparent that, with the exception of Black and Latin American men, there is no significant wage gap between visible minority and non-visible minority group membership for native-born workers. For all others, it is predominately among immigrants that the question of wage differentials arises for visible minority status. But because two of every three new immigrants to Canada is a visible minority, it is too easy to conflate disadvantage due to colour with disadvantage arising from the immigration circumstance. At a conceptual level, appropriate ameliorative effort would require different policy responses according as the disadvantage is associated with colour or immigration adjustment.
What implications do these findings have for public policy? Our essay attempts to disentangle some of the determinants of wages for visible minority Canadians. It is primarily an economic --- and not sociological nor anthropological --- examination of visible minority group membership. But at the very least, our findings should sound a warning that treating visible minorities as a homogeneous group for public policy, particularly employment equity, is ill advised. With more and more of Canada's immigrants being members of a visible minority group, and our evidence that the issue of hue and colour is predominately bound up with immigrant status, it is time to rethink Canada's emphasis for achieving equal opportunity in the labour market. Our findings suggest that the steps towards a colour-blind labour market offering opportunities for all Canadian workers may have to focus more towards helping immigrants adjust and integrate rather than the traditional prods embodying employment equity legislation. But the caveat to all this is the urgent need to examine the economic circumstances pertaining to native born Black men. Complacency that black-white differentials are more an issue for Americans than it is for Canadians should be forthrightly abandoned. US research has understandably framed the issue in terms of Black-White comparisons
for social and cultural reasons specific to their historical past. In Canada, we tend to approach the issue of race within a multi-ethnic and multicultural discourse, silently acknowledging the percolating issue of colour, but reluctant to privilege one non-white group over another. Yet, progression from economic and social marginalization towards full economic participation and social integration --- history shows --- has been markedly different for the Chinese and the Japanese, for the East Asians, and for the Blacks. The persistent disadvantage facing black men in Canada (and now, Latin American men) should no longer be submerged in a multicultural discourse nor confined exclusively within a visible minority context. The situation of native-born black Canadians can no longer be regarded as simply one extreme end of the variation that exists across all visible minority groups in Canada. The size of differential for native-born black men in Canada is too large, and too long lasting, to be viewed as a statistical outlier. Rather, the evidence is plain that economic disadvantage for blacks in Canada stems from unique structural features of Canadian society and economy, and it is hard to resist the suggestion that racial discrimination is an important factor. The statistically significant (and large) penalty for native born Latin American men is
also disturbing in this context. Will Latin American men, over time, imitate the trajectory of Chinese male immigrants? or that of native born Black men? These questions must be at the top of research agendas and policy priorities. Nonetheless, it bears repeating the necessary leitmotif of all enquiry concerning either race or immigrants in Canada --- visible minorities are an extremely heterogeneous category.
References Baker, M. and D. Benjamin. 1994. The Performance of Immigrants in the Canadian Labor market. Journal of Labor Economics 12(3): 369-405. Beach, C. and C. Worswick. 1993. Is There a Double-Negative Effect on the Earnings of Immigrant Women? Canadian Public Policy/Analyse de Politique 19(1): 36-53. Bloom, D. E., G. Grenier, and M. Gunderson. 1995. The changing labour market position of Canadian immigrants. Canadian Journal of Economics 28(4): 987-1005. Borjas, G. 1985. Assimilation, changes in cohort quality and the earnings of immigrants. Journal of Labor Economics 3: 463-89. Borjas, G. 1994. The Economics of Immigration. Journal of Economic Literature 32 (4): 1667-1717.
Carlson, L.A. and C. Schwartz. 1988. The Earnings of Women and Ethnic Minorities, 1959-1979. Industrial and Labor Relations Review 41(4):530-46. Chiswick, B. 1978. The effect of Americanization on the earnings of foreign-born men. Journal of Political Economy 86: 897-921. Christofides, L. N. and R. Swidinsky. 1994. Wage Determination by Gender and Visible Minority Status: Evidence from the 1989 LMAS. Canadian Public Policy/Analyse de Politique 20(1): 34-51. desilva, A. 1996. Discrimination Against Visible Minority Men. Hull: Human Resources Development Canada. Duleep, H. and M. Regets. 1992. Some Evidence on the Effects of Admissions Criteria on Immigrant Integration. In B. Chiswick (ed.) Immigration, Language and Ethnicity: Canada and the United States. Washington, D.C.: AEI Press.
Heckman, J. 1979. Sample Selection Bias as a Specification Error. Econometrica 47(1): 153-61. Hum, D. and W. Simpson. 1996. Canadians with Disabilities and the Labour Market. Canadian Public Policy/Analyse de Politiques 22(3): 285-299. Hum, D. and W. Simpson. 1999. Wage Opportunities for Visible Minorities in Canada. Canadian Public Policy/Analyse de Politique 25(3): 379-394. Hum, D. and W. Simpson. 2004a. Economic Integration of Immigrants to Canada: A Short Survey. Canadian Journal of Urban Research 13(1) 46-61. Hum, D. and W. Simpson. 2004b. Reinterpreting the performance of immigrant wages from panel data. Empirical Economics 29: 129-147. Kalbach, W., et al. 1993. Population Projections of Visible minority groups, Canada, Provinces and regions, 1991-2016.
Interdepartmental Working Group on Employment Equity Data, Statistics Canada, 1993. Levitan, S., G. Mangum, and R. Marshall. 1981. Human Resources and Labor Markets. New York: Harper & Row Li, P. 1993. Chinese Investment and Business in Canada: Ethnic Entrepreneurship Reconsidered. Pacific Affairs 66(2): 219-43. Pendakur, K. and R. Pendakur. 1998. The colour of money: earnings differentials among ethnic groups in Canada. Canadian Journal of Economics 31(3): 518-48. Schaafsma, J. and A. Sweetman. 2001. Immigrant earnings: Age at Immigration Matters. Canadian Journal of Economics 34 (4): 1066-1099. Shapiro, D. and M. Stelcner. 1997. Language and Earnings in Quebec: Trends over Twenty Years, 1970-1990. Canadian Public Policy/Analyse de Politique 23(2): 115-40.
Sowell, T. 1982. Weber and Bakke, and the presuppositions of Affirmative Action. In W.E. Block and M.A.Walker (eds.) Discrimination, Affirmative Action, and Equal Opportunity. Vancouver: The Fraser Institute, 36-63.
Table 1. Selected Characteristics of Canadians by Visible Minority (VM) Group Group Annual Earnings Hourly Wage Annual Hours Average Age % Male % Immigrant Non-VM $29,992 $19.43 1,474 43.91 50.94 10.56 VM $25,880 $17.89 1,433 41.92 47.57 86.07 Black $24,741 $16.93 1,416 41.95 41.80 79.00 Indo-Pakistani $26,855 $17.37 1,429 41.18 48.39 92.26 Chinese $25,913 $20.05 1,379 43.59 50.08 83.33 Non-Chinese Orientals a $28,244 $16.82 1,608 41.11 46.04 85.29 Arab $21,089 $18.13 1,198 41.60 47.48 88.00 Latin American $25,314 $15.99 1,545 39.27 52.08 96.47 a Non Chinese Orientals comprise Korean, Japanese, South East Asians, Filipinos, Oceanic. The sample size for each of these groups individually is relatively small. Note: Results are weighted by the cross-sectional weight to provide estimates for the Canadian population. Source: Survey of Labour and Income Dynamics (SLID) 2002 master (internal) file
Table 2. Sample Means for the Working Population in SLID Composite Hourly Wage $19.27 Log Hourly Wage 2.84 Visible Minorities? 10.24% Black? 1.60% Indo-Pakistani? 1.96% Chinese? 2.81% Non-Chinese Oriental? 2.44% Arab? 0.82% Latin American? 0.60% Immigrant? 16.21% Years since Migration 3.43 Age at Migration 3.60 Years of Schooling 13.64 Educated primarily outside Canada? 11.55% Years of work experience 17.79 Hours paid per week 36.82 Weeks worked per year 49.44 Male? 53.02% English as mother tongue? 52.34% French as mother tongue? 21.34% Reside in: Atlantic provinces? 7.70% Quebec? 23.93% Prairies? 16.43% British Columbia? 12.64% City over 500,000? 51.76% Rural area? 10.89% Sample size (weighted) 12,184,514 Sample size (unweighted) 26,759 Source: SLID, 2002
Table 3. Estimated Wage Gap Between non-visible Minority Members and Members of Visible Minority Group: 2002 vs. 1993 2002 MEN WOMEN Visible Minority Group Immigrant Native Born Immigrant Native Born All 12.7% *** 6.9% -0.5% -0.8% Black 22.2% *** 21.9% *** 2.8% 8.0% Indo-Pakistani 13.1% ** 6.3% 0.6% -26.7% *** Chinese 2.1% -1.8% a -6.0% 0.4% Non-Chinese Oriental 14.0% *** 6.2% -0.2% 5.9% Arab 13.7% * 6.0% -0.5% -22.5% Latin American 17.7% * 38.8% *** 3.1% -11.0% 1993 All 14.6% *** 3.1% 5.5% 2.7% Black 16.6% ** 25.6% *** -1.1% -13.0% Indo-Pakistani 19.0% *** -8.0% 2.3% 8.5% Chinese 17.3% ** -4.7% 9.9% -2.4% Non-Chinese Oriental 23.9% *** 1.6% 11.5% ** 17.2% Arab 11.2% -29.3%* -15.7% 23.4% Latin American -18.2% a 32.6%* 9.2% 9.3% a negative sign denotes wage difference in favour of the visible minority group Notes: *** denotes significance at the 1% level; ** significance at 5%; * significance at 10% Wage difference calculated as e ß -1 where ß is the coefficient from the log wage regression in Table 4. Source: Tables 4 (2002) and 5 (1993) in the Appendix.
APPENDIX Table 4. Regressions to explain Log Wage for Men and Women by Immigrant Status, 2002 MEN 2002 (1) (2) (3) (4) Immigrant Men Canadian Born Men Visible Minority? -0.136*** -0.071 (3.69) (1.08) Black? -0.251*** -0.247*** (3.95) (3.21) Indo-Pakistani? -0.140** -0.065 (2.03) (0.60) Chinese? -0.021 0.018 (0.42) (0.14) Non-Chinese Oriental? -0.151*** -0.064 (3.22) (0.54) Arab? -0.147* -0.062 (1.92) (0.18) Latin? -0.195* -0.491*** (1.77) (2.91) Yrs since migration 0.016*** 0.016*** (3.15) (3.24) Yrs since mig sqd -0.000-0.000 (0.15) (0.21) Age at migration 0.021*** 0.021*** (4.10) (4.36) Age at mig sqd -0.000*** -0.000*** (2.64) (2.97) Years schooling 0.043*** 0.042*** 0.052*** 0.052*** (9.52) (9.40) (26.25) (26.53) Educ ex Canada? -0.012-0.014-0.088-0.099 (0.25) (0.29) (0.65) (0.78) Yrs work experience 0.008 0.008 0.031*** 0.031*** (1.52) (1.52) (20.93) (20.99) Yrs work exp sqd -0.000-0.000-0.001*** -0.001*** (1.52) (1.55) (14.43) (14.49) Hours paid per week 0.001 0.001 0.001* 0.001* (0.61) (0.62) (1.93) (1.91) Weeks workd per yr 0.007*** 0.007*** 0.008*** 0.008*** (3.75) (3.68) (11.78) (11.80) English? -0.054-0.025 0.088*** 0.090*** (1.42) (0.63) (6.17) (6.47) French? 0.098 0.109 0.098*** 0.100*** (0.87) (1.00) (4.62) (4.78) Atlantic? -0.065-0.059-0.143*** -0.144*** (0.81) (0.75) (11.02) (11.12) Quebec? -0.085-0.065-0.076*** -0.077*** (1.45) (1.09) (3.66) (3.74) Prairies? -0.038-0.040-0.061*** -0.062*** (0.98) (1.04) (4.94) (5.01) British Columbia? 0.008-0.013 0.018 0.015 (0.18) (0.32) (1.03) (0.90) Cities > 500,000? -0.009-0.012 0.064*** 0.064*** (0.27) (0.34) (6.15) (6.15) Rural? -0.044-0.047-0.022* -0.022* (0.58) (0.64) (1.66) (1.65) Constant 1.501*** 1.496*** 1.371*** 1.375*** (9.29) (9.22) (30.59) (30.92) IMR -0.296*** -0.296*** 0.080*** 0.079*** (5.35) (5.23) (4.81) (4.82) Wald Χ 2 326.13*** 350.67*** 2235.20*** 2259.14*** Sample size 1,312 1,312 12,439 12,439
Table 4 (continued). Regressions to explain Log Wage by Immigrant Status, 2002 WOMEN 2002 (1) (2) (3) (4) Immigrant Women Canadian Born Women Visible Minority? 0.005 0.008 (0.15) (0.19) Black? -0.028-0.083 (0.45) (0.98) Indo-Pakistani? -0.006 0.237*** (0.11) (3.52) Chinese? 0.058-0.004 (1.18) (0.07) Non-Chinese Oriental? 0.002-0.061 (0.04) (0.84) Arab? 0.005 0.203 (0.05) (1.35) Latin? -0.031 0.104 (0.44) (1.24) Yrs since migration 0.019*** 0.019*** (4.23) (4.32) Yrs since mig sqd -0.000*** -0.000*** (2.90) (2.96) Age at migration 0.003 0.002 (0.53) (0.47) Age at mig sqd -0.000-0.000 (0.36) (0.36) Yrs schooling 0.042*** 0.043*** 0.061*** 0.061*** (10.34) (10.28) (24.81) (24.76) Educ ex Canada -0.062-0.061-0.147*** -0.129 (1.35) (1.34) (1.84) (1.45) Yrs work experience 0.005 0.004 0.017*** 0.017*** (1.02) (0.86) (9.36) (9.36) Yrs work exp sqd -0.000-0.000-0.000*** -0.000*** (0.17) (0.05) (3.55) (3.54) Hrs paid per week -0.000 0.000 0.000 0.000 (0.10) (0.03) (0.38) (0.41) Weeks worked per yr 0.006*** 0.006*** 0.008*** 0.008*** (4.51) (4.50) (15.00) (14.97) English? 0.046 0.059* 0.001 0.002 (1.39) (1.70) (0.05) (0.13) French? 0.050 0.057-0.008-0.008 (0.56) (0.63) (0.35) (0.32) Atlantic? -0.011-0.010-0.157*** -0.156*** (0.20) (0.17) (11.93) (11.89) Quebec? -0.049-0.038-0.075*** -0.074*** (0.82) (0.59) (3.31) (3.28) Prairies? -0.106*** -0.109*** -0.076*** -0.077*** (3.01) (3.08) (5.79) (5.84) British Columbia? 0.035 0.025 0.014 0.016 (0.91) (0.64) (0.85) (0.92) Cities > 500,000 0.068** 0.066** 0.101*** 0.101*** (2.21) (2.13) (8.92) (8.94) Rural? -0.126* -0.133** -0.016-0.016 (1.91) (1.99) (1.30) (1.30) Constant 1.464*** 1.460*** 1.278*** 1.279*** (11.44) (11.21) (23.50) (23.52) IMR -0.049-0.045 0.017 0.016 (0.68) (0.57) (0.62) (0.61) Wald Χ 2 266.08*** 269.97*** 1858.15*** 2270.98*** Sample size 1,239 1,239 11,769 11,769
Table 4 (continued). Regressions to explain Log Wage by Immigrant Status, 2002 Notes: Robust z statistics in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%; Regressions are weighted using the cross-sectional weights in SLID to reflect the Canadian population; Sample size reflects uncensored (reported) wage observations; Adjustment for sample selection bias derived from a sample selection model with schooling, experience, experience squared, regional dummies, gender, English and French, city size, visible minority, immigration status, years since migration, years since migration squared, age, family size, and other family income as explanatory variables. Source: SLID, 2002
Table 5. Regressions to explain Log Wage for Men and Women by Immigrant Status, 1993 MEN 1993 Variable Immigrant Men Canadian Born Men Intercept 1.769 (10.9) 1.797 (11.1) 1.678 (35.2) 1.677 (35.1) Visible Minority? -0.158 (3.5) -0.032 (0.7) Black? -0.181 (2.0) -0.296 (2.7) Indo-Pakistani? -0.211 (3.2) 0.077 (0.3) Chinese? -0.190 (2.5) 0.046 (0.6) NonChinese Orientals? -0.273 (3.6) -0.016 (0.2) Arab? -0.119 (1.2) 0.257 (1.7) Latin American? 0.167 (1.5) -0.394 (1.8) Yrs since migration 0.023 (4.1) 0.020 (3.6) Yrs since mig sqd -0.0004 (4.1) -0.0004 (3.5) Age at migration 0.012 (1.6) 0.014 (1.8) Age at mig squared -0.0002 (1.4) -0.0002 (1.6) Years of schooling 0.048 (10.9) 0.049 (11.2) 0.046 (27.0) 0.046 (26.9) Educ ex Canada? -0.078 (1.2) -0.097 (1.5) 0.080 (1.0) 0.068 (0.8) Years of experience 0.018 (4.5) 0.018 (4.3) 0.025 (31.3) 0.025 (31.1) Yrs of exp squared -0.0002 (4.8) -0.0002 (4.6) -0.0002 (30.1) -0.0002 (29.9) Hours paid per week -0.005 (2.8) -0.007 (3.4) -0.002 (3.2) -0.002 (3.2) Wks worked per year 0.002 (1.1) 0.002 (1.4) 0.004 (7.7) 0.004 (7.8) English? 0.067 (1.6) 0.067 (1.5) 0.054 (1.9) 0.061 (2.2) French? 0.235 (2.1) 0.233 (2.1) 0.042 (1.3) 0.049 (1.5) Atlantic? -0.153 (1.3) -0.137 (1.1) -0.109 (5.5) -0.111 (5.6) Québec? 0.041 (0.6) -0.051 (0.8) 0.027 (1.2) -0.028 (1.3) Prairies? -0.080 (1.6) -0.067 (1.4) -0.066 (4.1) -0.066 (4.1) B.C.? 0.052 (1.1) 0.068 (1.4) 0.074 (3.9) 0.072 (3.8) Cities > 500,000? -0.084 (2.0) -0.079 (1.9) 0.021 (1.6) 0.020 (1.6) Rural? -0.069 (1.0) -0.086 (1.2) -0.026 (1.8) -0.026 (1.9) R 2 0.376 0.392 0.259 0.261 F 16.497 13.819 127.480 96.461 Sample size weighted (unwtd) 679,527 (540) 679,527 (540) 3,787,808 (5,497) 3,787,808 (5,497)
Table 5 (continued). Regressions to explain Log Wage for Men and Women by Immigrant Status WOMEN 1993 Variable IM Women Cdn born Women Intercept 1.376 (9.5) 1.359 (9.3) 1.130 (24.0) 1.127 (23.8) Visible Minority? -0.057 (1.4) -0.027 (0.4) Black? 0.011 (0.1) 0.122 (0.9) Indo-Pakistani? -0.023 (0.3) -0.089 (0.5) Chinese? -0.104 (1.5) 0.024 (0.2) NonChinese Orientals? -0.122 (2.0) -0.189 (1.3) Arab? 0.146 (1.2) -0.267 (1.0) Latin American? -0.097 (0.9) -0.098 (0.3) Yrs since migration 0.016 (3.1) 0.015 (2.9) Yrs since mig sqd -0.0002 (2.4) -0.0002 (2.2) Age at migration -0.0016 (0.3) -0.0026 (0.4) Age at mig squared -0.0000 (0.2) 0.0000 (0.0) Years of schooling 0.060 (11.7) 0.061 (11.6) 0.068 (33.3) 0.068 (33.3) Educ ex Canada? -0.004 (0.1) 0.018 (0.3) 0.079 (0.9) 0.080 (0.9) Years of experience 0.018 (5.4) 0.018 (5.1) 0.018 (18.2) 0.018 (18.2) Yrs of exp squared -0.0002 (5.2) -0.0002 (5.1) -0.0002 (17.4) -0.0002 (17.5) Hours paid per week -0.006 (3.5) -0.006 (3.4) -0.001 (1.9) -0.001 (1.9) Wks worked per year 0.004 (2.1) 0.004 (2.4) 0.006 (12.4) 0.006 (12.4) English? 0.006 (0.2) -0.018 (0.4) 0.041 (1.4) 0.044 (1.5) French? 0.045 (0.4) 0.029 (0.3) -0.019 (0.5) -0.016 (0.5) Atlantic? 0.120 (1.0) 0.087 (0.7) -0.139 (6.4) -0.139 (6.4) Québec? -0.273 (4.2) -0.301 (4.5) -0.002 (0.1) -0.002 (0.1) Prairies? -0.135 (2.7) -0.126 (2.5) -0.086 (5.0) -0.086 (5.0) B.C.? -0.011 (0.2) -0.004 (0.1) 0.046 (2.4) 0.047 (2.4) Cities > 500,000? 0.087 (2.1) 0.087 (2.1) 0.093 (6.8) 0.093 (6.8) Rural? 0.006 (0.1) 0.010 (0.1) 0.000 (0.0) 0.000 (0.0) R 2 0.439 0.446 0.311 0.312 F 19.822 15.993 144.241 108.348 Sample size weighted (unwtd) 630,486 (502) 630,486 (502) 3,185,507 (4,805) 3,185,507 (4,805) Source: SLID, 1993 (internal master file)