« Julie Dixon's alternative to the bank tax | Main | Two Observations About Mexico I Cannot Entirely Explain »


Feed You can follow this conversation by subscribing to the comment feed for this post.

Typo: It's Simon Coulombe (Statistics Canada), not Serge Coulombe (U of Ottawa). (Thought Serge was getting even more eclectic in his research area!)

Yes. I think it's well worth breaking analysis down by sub-section and cross-section. And any policy response too. The simple "All effects are linear and additive" model clearly doesn't work. As your diagrams show. You can't just add the race dummy and the sex dummy.

(But why are certain ways of categorising people so salient, and others ignored? Wot about affirmative action for country folk who don't get as much education -- farmboys, like me?)

E.g.: why do Chinese women do better than white women, but Chinese men do worse than white men?

Hypothesis: More women entered the labour force. Take a standard Solow-Swan growth model, and this will cause a temporary fall in real wages (relative to LR trend), as the K/L ratio falls. In the long run, as higher income per capita (not per worker) causes higher savings, the steady state K/L ratio returns to its original level, and real wages return to their original level (relative to trend).

But women did not enter the labour force evenly. Many preferred (or were steered into) traditional "women's occupations". So wages in those occupations fell relative to traditional "men's occupations". http://www.statcan.gc.ca/bsolc/olc-cel/olc-cel?lang=eng&catno=11F0019M2007301

But maybe this is more true for some women than for others. Did Chinese women tend to be relatively more willing to go into "men's occupations"?

Actually, women's increased participation rate is surely the biggest macro story of the last 50(?) years. And the lagged effects of that change will take decades more to play out, both demographically, and in terms of K/L ratios, etc. Perhaps we macro should be paying more attention to it.

Nick, thanks for catching that typo!

Affirmative action for Canada's rural population is not an entirely stupid idea - living far from a university seems to make it more difficult to get an education.

Women in Canada is a great reference for any gender-based data. There's separate chapters on visible minority and immigrant women.

Some groups of visible minority women are more likely to study science and engineering than white women. I have two theories about this. One is that the traditional arts curriculum, with its heavy emphasis on European literature/knowledge, is less interesting to people of non-European heritage. Another theory - one especially relevant to people born outside of Canada (not the group studied by Hou and Coulombe)- is that if English is your second language and you've received an excellent foundation in math studying somewhere other than Canada, it makes sense to choose to study a subject where the English-writing demands are relatively low and the math demands are relatively high, like economics or engineering.

Do they lump native born and immigrant visible minorities together? Seems to me I've seen some work that suggests there's a significant difference there.

Jim, Canadian-born only - so they're looking at a pretty young cohort. There is a big difference between Canadian-born and immigrant outcomes. Hou and Coulombe also exclude a number of groups, e.g. Japanese Canadians who are one of the most successful demographics in Canada.

The number of Canadian born non-whites is very small, both as a percentage of all non-whites and a percentage of the labour force (A legacy of the white-focused immigration policy pre-67).

Up to now, the huge gap between white and non-white income (all else being equal) could have been attributed to cultural adaptation issues for immigrants. As the Canadian-born non-white cohort enters the labour force in large numbers, the nature of wage-differences by race in Canada will become easier to identify. The 2011 census should be especially useful.

Note that women make 70-90 percent of mens wages, all else being equal. So Chinese-Canadian women getting the same discount on a white man's wages as white women are. Non-white men are getting more or less the white woman's discount (depending on race).

Hou, Picot, Aydemir, Coulombe, and others at statscan have been doing all the interesting labour market work on this. Why no interest from the Academy?

I have a quite untested hypothesis to explain this sort of difference. I think it's mostly due to early founders of an immigrant group. If the first arrivals happen to be successful due to luck, skill, or whatever, they create opportunities for later immigrants who they feel connected to due to country of origin. Those early founders might well have differences in how they feel about women working, and those attitudes could have long-lasting and far-reaching effects.

NWIC - Some of the best Canadian work on visible minorities is being done by Krishna Pendakur at SFU. His brother, Ravi Pendakur has just taken up a position at U Ottawa. For a sample of Krishna's recent research, see here And of course there is Phil Oreopoulos's names research that received a lot of coverage in the media when it first came out. But there is some truth in what you say - I wouldn't be interested in doing a study like Hou and Coulombe's -- too straightforward.

Paul - there's a lot of interesting work being done on things like ethnic social capital, neighbourhood effects, and the role of culture. If you decompose the European Canadian population into ethnic sub-groups e.g. British, Greek, Italian you'll find fairly significant differences in average earnings there too.

I'm not sure why "straightforward" is a pejorative in the Academy. The work can start off straightforward, like some of Aydemir's early papers at Statscan. Buit that leads to the paper cowritten with Borjas on migration in Canada, US, and Mexico -- not strasightforward at all. Same for Hou's other papers. That's how research programmes work.

I've read Pendakur's papers (colour my world, etc -- cute title). Not bad, but less interesting and relevant than the work done at statscan ( Corak, Aydemir, Hou, Picot, Sweetman, Frenette, Morissette). Rietz and his students are always writing as well, but I don't like their approach.


While I agree that there are probably cultural adaptation issues for immigrants which may drive down wages, I'm not sure that you can assume that cultual adaptation issues go away for Canadian born workers. Part of the difficulty with these sorts of studies, is that they link wage differences to "race" (however defined), often with the implicit (or explicit) suggestion that any unexplained differences in wages between different racial groups is due to conscious or subconscious racial prejudice in the labour market.

Fair enough, if it's done properly. But the problem is that "race" is a tricky explanatory variable. First, it is often linked to other possible (but unobservable) explanatory variables. So, for example, race (at least as often defined) is likely correlated (in some cases, strongly) with culture. For example, the people who identify as in the above mentioned study as "Chinese", "Black" or "South-Asian" are likely not just racially "Chinese", "Black" or "South-Asian" but culturally "Chinese", "Black" or "South-Asian" (as an aside, is anyone struck by how artificial these "racial" groups are?). So it's impossible to tell to what extent the result is driven merely by skin colour (i.e., one's racial identity) or whether it's driven by cutlural attributes (i.e., one's cultural identity).

Now, one response to that is that, so what, that's a distinction without a difference. And I'd agree with that result, if you accept the premise that all cultures are equally "good" and so should produce identical results in the absence of discrimination. While I agree that all cultures are equally valid (in that people should be free to choose to live according to their own cultural attributes), I don't think you can accept the premise that they should produce identical results in the absence of discrimination. Culture matters, and probably goes a long way to explaining a lot of differences within broadly defined "racial" groups.

For example, as others have noted above, amongst "Whites", broadly defined, we see significant differences in labour market performances between, for example, Jews, Italians, and Portuguese (with Jews earning more than Italians, and Italians earning more than Portuguese). I suppose one could construct a model in which the labour market discriminates in favour of workers named "Goldstein" but against workers named "DaSilva" (but, curiously, not at all in respect of workers named Ianno), but that doesn't sound all that plausible. On the other hand, there are, for example, broadly recognized differences between those groups in terms of attitudes towards education (at least historically). Whereas Jews have traditionally been overachievers in obtaining education (especially at higher levels), Portuguese Canadians generally have not been (and indeed, at least within the Toronto Portuguese community, for many years, higher education was something that was not highly valued by parents).

And I suspect you could find similar examples in other "racial" groups (for example, had the above studied used as it's racial indicator East Asians, rather than "Chinese", I would expect to see similar differences between Koreans, Vietnamese, Japanese and Chinese (despite the fact that members of these groups are likely to be "racially" identical from the perspective of a would-be employer). And of course, if we see those differences WITHIN "racial" groups it's not entirely unreasonable to expect to see similar differences between different cultural groups in different "racial" groups. The key point, though, is that the diffences aren't driven either by race or by racism, but by differences in cultural attributes.

A second (though in some ways, related) problem with these studies of labour market disadvantage, is that while an apparent "racial" difference in incomes might be an indicator of racism in the labour market, it could also be an indicator of racism (or of some other disadvantage correlated with race) in other markets that affect labour market performance. For example, if you looked at the employment income of high school graduates in the US, you'd probably (in fact, almost certainly) see a significant gap between the performance of white and black high school graduates. That could be evidence of racial discrimination in the labour market. It could also be evidence that white high school graduates get better high school educations than black high school graduates and therefore are actually more productive. Certainly, that wouldn't be an unexpected result for anyone familiar with the quality (or lack thereof) of America's predominantly black inner-city schools. Unless you have some proxy for education quality (which is likely difficult to get in any readily measurable way), your indicator of labour market market discrimination is likely to be biased (so to speak).

And just to follow up on my last comment, those distinctions are important for policy purposes.

For example, if the apparent gap in wages between whites and non-whites is driven solely by race, there's a problem. Race is an inherently immutable aspect of one's identity (except, perhaps, for people who can "pass" as white or black or Asian, or what have you) while, if racism explains the gap, than query whether that racism can ever be erradicated (since, there are likely few, if any, societies in the world more "progressive" than Canada's). On the other hand, cultural is mutable. People can discard attributes of their culture that they don't like or which are disadvantageous, and adopt the attributes of others that they do (this process isn't instantaneous, however, so one would expect that it might take several generations). Moreover the value attributes to certain cultural attributes may vary with time (the high value given by Jews to education didn't do them much good in the middle-ages). To the extent that the differences in wages are driven by cultural attributes linked to productivity (and I should clariffy that I'm talking about differences in cultural attributes which are linked to productivity), than either those culture attributes (or the value given to them) will change (and wage gaps will disappear) or they won't, reflecting the choices of members of those cultures. In either case, it isn't a problem (unless we think that people have to adopt cultural attributes which maximize their market income - which is surely an unreasonable position).

On the other hand, if the wage gap is driven by not by racial discrimination in the labour market, but by racial discrimination or disadvantage in other areas, that calls for a different set of policy remedies than what is neccesary to address labour market disadvantage. To consider the example of the US, the focus of policy makers has been on labour market discrimination - which they've sought to address with affirmative action policies. Yet, those policies have not done much to address differences in labour market performance (despite ever broadening definitions of "discrimination"). At the same time, American policy makers have not taken meaningful steps to address the sharply declining quality of its, predominantly black, inner city public schools (both in absolute terms and relative to predominantly white suburban schools). As a result, Black teenagers (if they graduate from high school) come out of high school at a significant disadvantage relative to their White peers. The focus on purported labour market discrimination means that more serious causes of disadvantage (both in the labour market and in life generally) are ignored and resources are misallocated.

Also, either of these explanations might explain why the authors of this study didn't seen any difference in wages in the public sector. If public sector employers don't care about productivity (or can't measure it), or have hiring or promotion processes which don't reflect differences in productivity (affirmative action, labour unions), than differences in productivity linked to either cultural attributes or prior racial disadvantage aren't likely to be reflected in public sector wages.

Not White - other reasons why academics don't do studies like this one (a) academic success is tied to publications in "top" journals. It's much easier to get a top journal publication with US data. There are relatively few outlets for studies using Canadian data (b) US data is often easier to get than Canadian data. (c) Statistics Canada employees have an advantage when it comes to data access. Academics are limited to public use files of the census, unless we go down to a research data centre, but actually carrying out research at an RDC is not straightforward (if the only time you have free for research is 6:00 a.m. on Sunday morning or 9:30 p.m. on Thurs night they don't help, also it's hard to do research without coffee/internet access).

A couple of the people you give as examples of StatsCan researchers (Sweetman, Corak) are actually academics (Queens, U of O).

Though Corak I know worked at Statscan for years and would have access I think.

I was also going to suggest access to data as the big advantage and difference. It's almost a Catch 22 situation in my experience - you can't get access to the data unless you're part of the group, but you can't be part of the group unless you have access.

"Americans are now experimenting with affirmative action programs that target socio-economic disadvantage. Would it make sense for Canada to follow their lead?" -Frances

No, not beyond what we are already doing. To the best of my understanding, only two groups in Canada currently benefit from affirmative action programs: women and aboriginals.

I believe the focus of affirmative action going forward should be restricted to aboriginals and nobody else. Even there I would pray for great caution and reasonable restraint.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad