5 Some refinements

View the most recent version.

These findings all pertain to averages, speaking to the overall patterns in the country. However, the large sample size of the census allows us to explore the variation in the data in more detail than possible with any other data source. Figures 1 and 2, for example, offer scatter plots of parent and child years of schooling used in the regression analysis presented in the first column of Tables 6 and 7. This illustrates the grouped information from each of the 70 countries making up our analytical file. A picture of this sort would be available from the Ethnic Diversity Survey but, because of the smaller sample size, for only about 30 countries. The weighted least squares regression line with slopes of 0.136 for father-son years of schooling, and 0.102 for father- daughter years of schooling are included in the figures, as are the average years of schooling for Canadian-born fathers and their Canadian-born children. These latter points are for illustrative purposes and are not used in the regression.¹¹

The figures make clear that the children of immigrants are more educated than their counterparts with Canadian-born parents: the educational attainment of the latter lie below the level predicted by the regression line for the number of years of schooling their fathers had on average. While there is rapid regression to the mean among immigrant children—much more rapid than for the native Canadian population—it should be clear that this mean is an immigrant-based mean: the children of immigrants are regressing to a different, higher average than the children of Canadian-born parents.

The figures also make clear that in spite of the general tendencies there is a good deal of variation about the estimated regression lines. For the strong majority of countries, immigrant fathers have more than the average education of Canadian-born fathers, and this advantage is passed on to the next generation, both sons and daughters having more years of schooling than their Canadian-born cohort with Canadian-born parents. It is only the case in four countries that immigrant educational attainment is below the Canadian average and that this gap continues to be reflected in the next generation of sons and daughters. For all the other countries in which father's education is less than the average, the children make relative gains and exceed the Canadian average.

We use the information in these figures combined with similar information on weekly earnings in Aydemir, Chen and Corak (forthcoming) to highlight particular communities of concern. One could imagine that matters of integration into the broader community would be particularly salient under at least the following two circumstances. First, consider a situation in which fathers come to the country with greater than average schooling, yet earn less than the average. This might in part be the situation currently playing itself out because of the reported difficulties in having foreign education credentials recognized in the Canadian labour market. If these fathers then witness a similar scenario occurring for their children, it can reasonably be imagined that a sense of frustration or lack of belonging to the host country could develop both among them and among their adult sons and daughters. They may be willing to shoulder the costs of below average earnings in spite of having above average education, but to see that these costs do not lead to an improved situation for their children may change their perspective and that of their children. The second particularly pertinent scenario might be one of intergenerational transmission of lower earnings and education: fathers who come to the country with below average education and below average earnings see their children to grow up to also have lower education and earnings than the mainstream. This is a scenario in which social exclusion may be felt across generations.

In order to illustrate the relevance of these two possibilities we cross-classify the information in Figures 1 and 2 with similar information on parent-child weekly earnings. The results are offered in Tables 11 and 12, panels 1 looking at countries of origin in which fathers have on average more years of schooling than their Canadian counterparts, and panels 2 in which they have less. Focusing for the moment on Table 11, dealing with the father-son relationship, it is notable, firstly, that there are no cases of downward education mobility: if fathers have above average education, so do the sons. Second, in 57 of 70 countries the fathers have above average education, and in two thirds of these they also have below average earnings. But in 11 of these 38 cases the sons go on to earn less than the Canadian average, in spite of having above average education. So only a very small number of countries representing a small total population occupy this potentially challenging position. That said, these are dominated by the Caribbean countries and with the addition of West Africa likely represent a visible minority group highlighted by Reitz and Bannerjee (2007), and is the basis for their rather pessimistic perspective on the integration of second-generation immigrants.¹²

To address our second scenario, there are only 12 countries in which fathers have less than average years of schooling and less than average earnings, and in all but two of these cases the sons go on to have above average earnings, breaking out of this potentially challenging starting point. Some of these sons do this by having above average education (6 of 10), while others continue to have below average education in spite of having higher earnings. There are only two cases of an intergenerational transmission of disadvantage in earnings, though not of education: Cyprus and Greece, and neither of these countries have been highlighted by the existing literature as being particular flashpoints for discontent.

The situation for daughters, presented in Table 12, is in fact quite different than for sons. There is only one case of downward education mobility, Norway, and only one case in which fathers with above average education and below average earnings have daughters who grow up to also have above average education and below average earnings. For 37 of the 38 countries of origin with fathers in this situation the daughters go on to have both higher education and higher earnings than their Canadian counterparts. Furthermore, there are no examples of the intergenerational transmission of lower education and earnings. There are 13 countries of origins in which fathers on average have less education than the Canadian average, and in 12 of these they have below average earnings. But there is only one case in which the daughters find themselves with below average education and earnings.

¹¹ We examined the robustness of the regression estimates by successively dropping a single observation from the estimation and recalculating the slope before then re-including it in the estimation and dropping the next observation. There are no particularly influential observations in the data, the estimated slope not changing at all. The only exceptions to this—for both sons and daughters—were the estimates excluding the United Kingdom, Italy and Portugal. Without the United Kingdom, the slope for fathers-sons is just over 0.16, and similarly without Italy; while without Portugal it is 0.11. These are all within one standard error (0.038) of the original estimate of 0.136. The general patterns and conclusion also hold for the father-daughter analysis. Also, to be strictly correct, the averages depicted in the figures are for the reference case of someone living in Ontario.

¹² The fact that Japan is also included in this group might be an anomaly. A closer look at the data shows that with respect to weekly earnings the data for both fathers and sons are essentially the same as the Canadian average, differing by less than 0.01 log points. Japan could just as easily be included as above the average, as below.