Skip to main content

Canadian Pacific Rail workers inspect tracks at the company's Port Coquitlam yard east of Vancouver, on Tuesday Dec. 4, 2012.

The debate over whether Canada is experiencing a jobs shortage, a skills shortage, both or neither has heated up in recent months – fuelled in no small part not only by the stalling economic recovery, but also by a few questionable comments from the recently minted Employment Minister, Jason Kenney.

One such comment suggested critics of Mr. Kenney's skills-shortage view didn't understand the issue because they weren't looking at the data by region and industry. The Minister neglected to provide reliable data to support his view, for good reason. There isn't any. While Statistics Canada recently began collecting job vacancy statistics, the data aren't terribly useful, and the Canadian survey's shortcomings are apparent when contrasted with its U.S. equivalent.

To address a glaring gap in Canadian labour market data dating back to the 1970s, Statistics Canada in 2011 began to ask employers about job vacancies. Combining data from its Business Payroll Survey (BPS) and Labour Force Survey (LFS), Statscan produces monthly Job Vacancy Statistics (JVS) reports, published with a three-month lag (for example, the most recent data, published Tuesday, relates to September). In the U.S., the Bureau of Labor Statistics Job Openings and Labor Turnover Survey (JOLTS) began in 2000, preceded by several similar U.S. labour turnover surveys dating back to 1959. It is conducted separately from the BLS payroll survey; its monthly job vacancy report has just a one-month lag.

The first difference between the surveys is their scope. Statscan's JVS sample excludes nearly a quarter of employer businesses, among them defence and federal, provincial and territorial public administration. The JOLTS sample does not exclude these employers.

Statscan's JVS has a relatively larger sample size, which should make its data more reliable. But in reality, it makes little difference at the sub-national level; nearly all the JVS data at the provincial level is suppressed due to data quality and confidentiality. Part of the problem is that, as a result of non-responses from surveyed businesses, Statscan has had to "impute" (in essence, guesstimate) 30 per cent of the data using other businesses' responses – curious for a supposedly mandatory survey.

Both Statscan's JVS and JOLTS produce statistics on job vacancies by industry at the national level. Neither, however, collect data on job vacancies by occupation – critical in determining what, if any, role job-skills mismatch plays in persistent Canadian labour market slack. Apparently Statscan's previous version of the JVS, conducted in the 1970s, inquired about occupation.

Questionnaire content is where Statscan's JVS really comes up short. It's essentially one additional two-part question to employers: Did their business have any vacant positions at the end of the reference month, and if so, how many. Statscan advises respondent businesses that a vacancy can be "full-time, part-time, permanent, temporary, seasonal, on call, etc."

While the JOLTS questionnaire is also essentially two questions, they are detailed. Importantly, JOLTS requests information not only on total employment and job openings, but also on hires, recalls, quits, layoffs, discharges and other separations (including retirement).

A recent Federal Reserve Bank of San Francisco paper explains: "The fact that JOLTS includes both data on vacancies and hires is very useful, since it provides direct evidence on the rate with which potential employers and employees are matched in the labour market. This measure is the number of hires per vacancy, known as the vacancy yield."

Looking at the JVS data for the Employment Minister's home province illustrates the significance of vacancy yield. For September, half the Alberta job vacancies Statscan was confident enough to release (marked "Use with caution") were in Retail, Food, Accommodation and Other services, up slightly from September, 2012. These service sectors primarily offer precarious employment. Without knowing the vacancy yield, it's impossible to tell whether employers in these service sectors are listing more vacancies in anticipation of increased demand, re-filling high turnover, or simply carrying forward unfilled vacancies.

The lack of reliable job vacancy data at the regional level even by broad industry aggregate, as well as the complete lack of occupation and labour turnover data, renders the JVS in its current form incredibly limited.

The solutions are obvious: Increased sample size and better follow-up to increase reliability, and the addition of occupation and turnover questions to provide useful insight. The arguments against implementing these, or any, changes are well-worn: Increased cost, complexity and respondent burden.

If, as Mr. Kenney suggests, critics need to look at the job vacancy data by region and industry to understand whether/where skills shortages are contributing to persistent labour market slack, it should be incumbent on the government to ensure such data are at least available, along with occupation and turnover data to make them useful.

Sam Boshra is an independent Montreal-based economist, and editor for Angella MacEwen is a senior economist with the Canadian Labour Congress and research fellow at the Broadbent Institute.

Interact with The Globe