Categories
Employment Immigration Race and ethnicity

March 2014 LFS: An alternative view. The R word, illustrated (updated April 25, 2014)

Chart 1a Unemployment by immigration and visible minority statusImm_VM_unemp_2006

Source: Census of Population, 2006 [Canada] Public Use Microdata File, Individual File: Canada. 2.7% sample, Statistics Canada

The March 2014 Labour Force Survey (LFS) release reported a rare increase in employment greater than the error term; unlike recent months, that change wasn’t attributable to self-employment, either. Rather, it was part-time, public sector employment. Despite the less-than-encouraging news, Canadian stocks rallied, underscoring the disconnect between capital market and real economy.

Now on to the months-long tradition of ignoring the monthly numbers and discussing broader trends and issues. This month, it’s the R word.

Immigrant LMI still MIA

The March 2014 LFS release marks four months down, eight to go on StatsCan’s anticipated reporting of immigrant labour market information (‘some time this year’). The next (April 2014)  release will mark the 100-month anniversary since the LFS began collecting (January 2006) but not reporting immigrant employment data.

For economists and other interested social researchers, what’s anticipated this year is only immigrants being mentioned in the monthly report. According to StatsCan, the LFS Public Use Microdata File (PUMF) will not be updated to include immigration variables until at least 2018. Until then, those interested in any immigration-related cross-tabs from the LFS data will be required to pay StatsCan for ‘custom’ tabulations. There’s an obvious moral hazard when a national statistical agency can financially gain from withholding data from the public, but that’s a subject to be covered at a later date…

The R word 

The lack of immigrant LMI was mentioned in a previous Economy Lab post. One commenter (see, they are read) pointed out the issue isn’t just one of immigration status, but broader labour market discrimination faced by racialised minority groups. Which is absolutely true.

There are a couple of obvious issues with using immigrant status as a proxy for race: As Chart 1 clearly illustrates, not all visible minorities are immigrants, nor are all immigrants visible minorities. First-generation Canadians from countries with fairly diverse populations where economic outcomes there may also be affected by racial/ethnic discrimination – like the US and UK, for example – may find themselves facing similar discrimination upon immigration to Canada.

The 2006 long-form Census data, illustrated in Chart 1, underscores these points:
– Canadian-born vismin comprised nearly a third (30.2%) of all vismin.
– Immigrant non-vismin comprised nearly half (45.6%) of all immigrants.
– Immigrant non-vismin had lower unemployment (5.1%) than immigrant vismin (8.6%, aggregate) as well as Canadian-born non-vismin (6.3%).

Problem: While a number of developed market economies with large and diverse populations and workforces – like the US and UK, for example – have national statistical agencies that regularly ask about race/ethnicity while surveying their populations’ labour market outcomes, Canada isn’t one of them.

Why not?

The question was forwarded to StatsCan – on March 20, 2014. The very Bureaucratese (it’s a language) reply received from StatsCan – on April 2, 2014:

… a number of variables were considered by Statistics Canada in collaboration with HRSDC and Immigration Canada for a small update to the LFS content. The list included the current set of LFS immigrant variables as well as visible minority, among others. The assessment of these variables for possible inclusion had to take into account: a) content space available on the questionnaire (i.e., there was a need to drop existing variables to make room for any new ones); b) respondent burden; c) review by STC experts; d) discussions with major stakeholders. The final decision based on the assessment and the removal of previously existing variables from the LFS, was that the current set of immigration variables be added.

Apparently asking both questions would break the Canadian LFS – despite the US Bureau of Labor Statistics and UK Office for National Statistics managing to include both country of birth and race/ethnicity in their respective labour force surveys without incident. The StatsCan/HRSDC/CIC ‘experts’ decided that immigrant status alone was sufficient.

(For readers wondering how many of the so-called ‘experts’/stakeholders involved in the decision to exclude race/ethnicity from the Canadian LFS were vismin themselves, feel free to write the respective federal agencies and ask… and let us know how they respond.)

Backward, and backsliding

Questions on race/ethnicity are/were included in other StatsCan surveys, for example the Survey of Labour and Income Dynamics (SLID). That survey used the LFS sampling frame; unfortunately, it was a voluntary survey with a small sample size, making it less than reliable when it came to measuring characteristics of certain subpopulations – like visible minorities.

The SLID was discontinued with the 2012 final release. It was replaced by the Canadian Income Survey (CIS), a supplement to the LFS surveying a small subsample on a voluntary basis. It’s unclear at this point whether CIS queries race/ethnicity, as the questionnaire isn’t publicly available.

The Canadian Survey of Economic Well-being (CSEW), another household survey recently introduced by StatsCan with nearly the same characteristics as CIS (LFS supplement, small subsample, voluntary), does not ask about race/ethnicity.

Most of the research on race/ethnicity and socio-economic outcomes in Canada has historically relied on the long-form Census, the only mandatory national household survey that asked the question. Unfortunately, that was discontinued and replaced by the voluntary National Household Survey (NHS) in 2011. Seeing as StatsCan’s already conceded the 2011 NHS immigration data is unreliable, it’s a safe bet the vismin data likewise is.

There are other indicators of socio-economic well-being beyond employment and income – health and legal justice outcomes, for example. As mentioned in reference to Rob Ford not even being charged let alone tried/convicted after openly admitting to an offence under the Criminal Code, Canadian crime and justice stats are also not readily made available by race – although one can try to get them the hard way.

Wilful ignorance going forward

The 2011 NHS found a fiftth of Canadians self-identified as visible minority. Previous Census-based StatsCan and CIC projections had vismin comprising up to a quarter of Canada’s population by as early as 2017 and a third of its population by 2031.

Instead of choosing to collect more and better quality data on the socio-economic well-being of a demographic comprising a large and growing share of its population, Canada’s statistical, employment and immigration authorities – effectively, its government – has decided less is more.

Editor’s note
The referenced issue is not a simple matter of political party affiliation (in response to some readers mistakenly taking the current government to task). It should be noted the referenced StatsCan/HRSDC/CIC decision to exclude race/ethnicity from the LFS dates back to (at least) the mid-90’s.

Update (25/04/2014)
Charts, related comments added,

Chart 1b Unemployment by visible minority status and place of birthVM_Imm_unemp_2006

Source: Census of Population, 2006 [Canada] Public Use Microdata File, Individual File: Canada. 2.7% sample, Statistics Canada

Leave a Reply

Your email address will not be published. Required fields are marked *