I believe that you are reading the Statistics Canada table incorrectly, though the way the table is presented is not helpful. You quote the number $50,280 for the year 2021. That number comes from a row titled "Total - Visible minority". The punctuation mark looks like a dash, but I think it might be a minus sign. I do not understand what it means in either case, but I don't think it's the average for all visible minorities.
However, the row below it is titled "Total visible minority population", and it contains the number $43,920. That seems to be the row containing the average you want. The bottom row, "Not a visible minority", contains $52,550. The ratio of those two numbers is 83.6%. I consider that a plausible estimate of how little visible minorities are earning, while 95.6% (the number you cite) seems too high an estimate.
Using those two rows, the ratios for 2006 through 2021 are 81.4%, 86.9%, 81.2%, and 83.6%. On the one hand, this looks like 2.2% progress from 2006 to 2021; on the other hand, the numbers go up and down so much that I suspect we're looking at noise rather than signal.
I find it peculiar that there are no categories in the table for indigenous minorities. I was able to find the questions for the 2021 census online at https://www.statcan.gc.ca/en/statistical-programs/instrument/3901_Q2_V6. There is a list of categories such as "White" and "Chinese" as question 25, while indigenous categories are handled separately as questions 26 through 29. This may have made sense as a strategy for gathering data, but it resulted in some unacceptable gaps in the way data was summarized.
Nothing I write should be taken as disagreeing with your main point, that people in visible minorities are not being paid as much as people who are not in those minorities. However, the data available don't make it straightforward to see how much progress we're making in alleviating that situation.
I **really** appreciate readers checking my math - because I definitely make mistakes. However I suspect that this time I my numbers are correct. The Statistics Canada page I linked to allows us to edit the parameters. Here are the changes I made:
* Under the Employment income tab, I specified only "Average employment income"
* Under the Visible minority tab, I selected just the "Total - Visible Minority" and "Not a visible minority" boxes
* In the Census year tab I selected all the available years
The problem isn't the math. The problem is which numbers were selected from a confusing table to become part of the math.
There is one row titled "Total - Visible minority" and another titled "Total visible minority population". The second is indented a bit and has the same indentation as the bottom row, labeled "Not a visible minority". That makes me suspect that the first one really means "Total, for all of the people for whom the visible minority question was answered".
If you look carefully at the numbers, the value of $50,280 for the first row is higher than all but two of the individual visible minority categories, Chinese at $50,302 and Japanese at $51,450. That's what made me suspect that it wasn't really the average for all people who are in visible minorities, even though I know that a large proportion of them are in fact Chinese. Just eyeballing the numbers, and not knowing the proportions, one would think that the weighted average is somewhere in the mid $40K range, and that's what the second row shows at $43,920. That's why I think it's the second row, not the first, that actually contains the average earnings for the visible minority population.
The confusion ultimately stems from a very bad naming decision on Statistics Canada's part. They called the dimension "Visible minority" but also called a collection of categories within that dimension "visible minority". If they had called the overall dimension something like "Racialization", then there could have rows titled "Total - Racialization", "Total visible minority", and "Not a visible minority", which would have been much clearer.
You're definitely correct that Statistics Canada's naming schema here is awful. And, based on my previous experience, contacting them for clarity is probably a waste of time.
But I'm not convinced that "Total - Visible minority" is a good fit with "Total, for all of the people for whom the visible minority question was answered" - in fact, I can't remember if the census had a simple "visible minority/not visible minority" question on which it could reasonably be based.
Having said that, I can't rule it out. But I would note that your figures (ranging between *disparities* of 19% and 13% through those four census years) seem way too high: could there really be such a vast pay gap across the entire population?
Yes, there could. If the gap were only 5% or so, we wouldn't notice such a large disparity (in society, not in the table) between the majority and the visible minority groups.
I did a brief web search for other articles about racial wage gaps, and I found one from the Conference Board of Canada at https://www.conferenceboard.ca/hcp/racial-gap-aspx/, looking only at some 2010 data, and limited to university-educated, full-time employees. The top figure shows gaps of over 13% in the four most populous provinces, which comprised 86% of the population in 2011, and over 17% in Ontario plus Quebec, which comprised 61%.
So, these two sources suggest that there is an earnings gap well above the 5% suggested by the original Substack post.
Finally, even if the census does not have a "visible minority/not" question, that information can be derived by simply combining all of the named minority categories, which is what appears to have been done.
I don't have any past experience customizing a Statistics Canada table via the web interface. Is there a way to simply get the number or proportion of responses to the income question by racial category? If there is, then one should be able to quickly see which rows are sums of which other rows, and that would help to disambiguate the row labels.
I just asked ChatGPT for its opinion. It wasn't 100% sure ("The title of this column could be somewhat misleading if it's intended to mean the same as the previous column. However, given the different median income value ($32,000), it suggests a different grouping or methodology might be in play."), but GPT suggested that the "Total visible minority population" data might include "all individuals identified as belonging to a visible minority (including those not employed), thus perhaps lowering the median income figure when compared to just those visibly in minority and employed (reflected in the first column)."
If that's true (and it seems reasonable), then the choice of which value is the better comparison to the "Not a visible minority" numbers would depend on whether "Not a visible minority" represents all members of that class or just those who were employed. My guess is that it's the former.
I have now found a StatsCan table that shows exactly what the ambiguous row labels mean. The web page at https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=9810035101 defaults to showing simple counts of population in various visible minority classifications, extrapolated from the 25% sample. In that table, the top row of data contains total population counts for Canada. Here are three column labels and the corresponding counts:
"Total - Visible minority": 36,328,475.
"Total visible minority population": 9,639,200
"Not a visible minority": 26,689,275
So, the "Total" label that contains a hyphen refers to the entire population, visible minority or otherwise. To look at the data for just the visible minorities, one must use the "Total" label without the hyphen.
Like so much of what passes for regulation, large corporations whose inherent bureaucracies can more easily navigate, let alone cheerlead, such intrusions, "encourage" the state to inflict the same on existing and more importantly, potentially competitive firms thus protecting themselves while advancing their cultural Marxist P.R. bona fides.
I believe that you are reading the Statistics Canada table incorrectly, though the way the table is presented is not helpful. You quote the number $50,280 for the year 2021. That number comes from a row titled "Total - Visible minority". The punctuation mark looks like a dash, but I think it might be a minus sign. I do not understand what it means in either case, but I don't think it's the average for all visible minorities.
However, the row below it is titled "Total visible minority population", and it contains the number $43,920. That seems to be the row containing the average you want. The bottom row, "Not a visible minority", contains $52,550. The ratio of those two numbers is 83.6%. I consider that a plausible estimate of how little visible minorities are earning, while 95.6% (the number you cite) seems too high an estimate.
Using those two rows, the ratios for 2006 through 2021 are 81.4%, 86.9%, 81.2%, and 83.6%. On the one hand, this looks like 2.2% progress from 2006 to 2021; on the other hand, the numbers go up and down so much that I suspect we're looking at noise rather than signal.
I find it peculiar that there are no categories in the table for indigenous minorities. I was able to find the questions for the 2021 census online at https://www.statcan.gc.ca/en/statistical-programs/instrument/3901_Q2_V6. There is a list of categories such as "White" and "Chinese" as question 25, while indigenous categories are handled separately as questions 26 through 29. This may have made sense as a strategy for gathering data, but it resulted in some unacceptable gaps in the way data was summarized.
Nothing I write should be taken as disagreeing with your main point, that people in visible minorities are not being paid as much as people who are not in those minorities. However, the data available don't make it straightforward to see how much progress we're making in alleviating that situation.
Hi,
I **really** appreciate readers checking my math - because I definitely make mistakes. However I suspect that this time I my numbers are correct. The Statistics Canada page I linked to allows us to edit the parameters. Here are the changes I made:
* Under the Employment income tab, I specified only "Average employment income"
* Under the Visible minority tab, I selected just the "Total - Visible Minority" and "Not a visible minority" boxes
* In the Census year tab I selected all the available years
I just ran that again and I got the same numbers.
The problem isn't the math. The problem is which numbers were selected from a confusing table to become part of the math.
There is one row titled "Total - Visible minority" and another titled "Total visible minority population". The second is indented a bit and has the same indentation as the bottom row, labeled "Not a visible minority". That makes me suspect that the first one really means "Total, for all of the people for whom the visible minority question was answered".
If you look carefully at the numbers, the value of $50,280 for the first row is higher than all but two of the individual visible minority categories, Chinese at $50,302 and Japanese at $51,450. That's what made me suspect that it wasn't really the average for all people who are in visible minorities, even though I know that a large proportion of them are in fact Chinese. Just eyeballing the numbers, and not knowing the proportions, one would think that the weighted average is somewhere in the mid $40K range, and that's what the second row shows at $43,920. That's why I think it's the second row, not the first, that actually contains the average earnings for the visible minority population.
The confusion ultimately stems from a very bad naming decision on Statistics Canada's part. They called the dimension "Visible minority" but also called a collection of categories within that dimension "visible minority". If they had called the overall dimension something like "Racialization", then there could have rows titled "Total - Racialization", "Total visible minority", and "Not a visible minority", which would have been much clearer.
Specifically: using your "Total - Visible minority" numbers, here's what the disparities between minority and non-minority income would look like:
2006 - 18.63%
2011 - 13.10%
2016 - 18.78%
2021 - 16.42%
You're definitely correct that Statistics Canada's naming schema here is awful. And, based on my previous experience, contacting them for clarity is probably a waste of time.
But I'm not convinced that "Total - Visible minority" is a good fit with "Total, for all of the people for whom the visible minority question was answered" - in fact, I can't remember if the census had a simple "visible minority/not visible minority" question on which it could reasonably be based.
Having said that, I can't rule it out. But I would note that your figures (ranging between *disparities* of 19% and 13% through those four census years) seem way too high: could there really be such a vast pay gap across the entire population?
Yes, there could. If the gap were only 5% or so, we wouldn't notice such a large disparity (in society, not in the table) between the majority and the visible minority groups.
I did a brief web search for other articles about racial wage gaps, and I found one from the Conference Board of Canada at https://www.conferenceboard.ca/hcp/racial-gap-aspx/, looking only at some 2010 data, and limited to university-educated, full-time employees. The top figure shows gaps of over 13% in the four most populous provinces, which comprised 86% of the population in 2011, and over 17% in Ontario plus Quebec, which comprised 61%.
That web page references a report titled "Canada's Colour Coded Labour Market", by Sheila Block and Grace-Edward Galabuzi, based on 2006 census data. (Its URL is https://www.policyalternatives.ca/sites/default/files/uploads/publications/National%20Office/2011/03/Colour%20Coded%20Labour%20Market.pdf.) In the Executive Summary, that report says "Racialized Canadians earn only 81.4 cents for every dollar paid to non-racialized Canadians."
So, these two sources suggest that there is an earnings gap well above the 5% suggested by the original Substack post.
Finally, even if the census does not have a "visible minority/not" question, that information can be derived by simply combining all of the named minority categories, which is what appears to have been done.
I don't have any past experience customizing a Statistics Canada table via the web interface. Is there a way to simply get the number or proportion of responses to the income question by racial category? If there is, then one should be able to quickly see which rows are sums of which other rows, and that would help to disambiguate the row labels.
Those are useful data points, but not conclusive.
I just asked ChatGPT for its opinion. It wasn't 100% sure ("The title of this column could be somewhat misleading if it's intended to mean the same as the previous column. However, given the different median income value ($32,000), it suggests a different grouping or methodology might be in play."), but GPT suggested that the "Total visible minority population" data might include "all individuals identified as belonging to a visible minority (including those not employed), thus perhaps lowering the median income figure when compared to just those visibly in minority and employed (reflected in the first column)."
If that's true (and it seems reasonable), then the choice of which value is the better comparison to the "Not a visible minority" numbers would depend on whether "Not a visible minority" represents all members of that class or just those who were employed. My guess is that it's the former.
I have now found a StatsCan table that shows exactly what the ambiguous row labels mean. The web page at https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=9810035101 defaults to showing simple counts of population in various visible minority classifications, extrapolated from the 25% sample. In that table, the top row of data contains total population counts for Canada. Here are three column labels and the corresponding counts:
"Total - Visible minority": 36,328,475.
"Total visible minority population": 9,639,200
"Not a visible minority": 26,689,275
So, the "Total" label that contains a hyphen refers to the entire population, visible minority or otherwise. To look at the data for just the visible minorities, one must use the "Total" label without the hyphen.
Like so much of what passes for regulation, large corporations whose inherent bureaucracies can more easily navigate, let alone cheerlead, such intrusions, "encourage" the state to inflict the same on existing and more importantly, potentially competitive firms thus protecting themselves while advancing their cultural Marxist P.R. bona fides.