Temporary foreign workers. Unpaid interns. Current jobless rates on aboriginal reserves.
These are just a few examples of the known unknowns in Canada's labour market – statistical gaps in our knowledge about how people are faring in the jobs market.
Statistics Canada is in the midst of a redesign of its labour force survey (LFS), the first phase of which will be completed next year, that aims at ensuring its "relevance and quality."
Reliable, complete and up-to-date labour market data is a crucial component of government policy, influencing everything from educational priorities to immigration.
Yet, fallout from faulty or missing labour market information has made headlines on a number of issues this year alone. It's been hard to pinpoint the size of the temporary foreign worker program (The Globe and Mail has reported that some employers say the tally for their organizations is miscounted).
Some TFWs are working on First Nations reserves with high unemployment rates – but it is unclear just how high jobless rates are on reserves. And earlier this year, the Finance Department had to correct its estimates of job vacancies when it emerged it was relying on a software program that was skewed by including Kijiji listings.
Statscan itself has faced budget cuts and staff reductions, while a memo has shown Employment and Social Development Canada has cut spending on labour market information by more than 20 per cent over the past two years.
The Conservative government said in June it will now spend an additional $14-million a year on two Statscan surveys to produce detailed regional jobs data. The statistical agency says it plans to release an expanded job-vacancy survey next August, along with annual wage data, starting in 2016.
As Statscan releases its labour force survey Friday, The Globe asked several economists and people who analyze the jobs market what additional labour market information would be useful, and why. They were also given the option of saying nothing needs to be added. Here are their ideas:
Improve job vacancy survey and survey on skills shortages
It would be helpful to have better labour demand data. Job vacancies are now reported as a three-month moving average and released three months after the surveyed months. Having better labour demand data would allow comparison between labour supply and labour demand. It would be nice to have this data broken down into census metropolitan area to examine regional labour demand. – Parliamentary Budget Office
We still need sub-provincial information and finer occupational detail on job vacancies. It helps nobody to know, for example, there is an engineering job (without specifying what type of engineering) somewhere in Ontario. – Don Drummond, former Toronto Dominion Bank chief economist and chair of the Advisory Panel on Labour Market Information
Look to the U.S. – publish more detailed surveys more quickly
Statscan's survey of employment, payrolls and hours is published two months after the labour force survey. Instead, time the LFS and the SEPH to publish them together on the same release day for the same month. We should also have faster employment insurance claims data, and they should be weekly, not just lagging and monthly. We should produce fresh job vacancies data like JOLTS (job openings and labour turnover survey) in the U.S. and other measures in that survey like job quit rates.– Derek Holt, economist at Bank of Nova Scotia
Improve survey on employment and income outcomes of graduates by discipline
This information could be captured by linking student identification numbers to income tax returns (anonymously). We now "only get a snapshot every five years on outcomes two years after graduation. Young people need this information to guide their decisions on future studies. Education institutions should use the information to allocate resources." – Don Drummond, former Toronto Dominion Bank chief economist and chair of the Advisory Panel on Labour Market Information
Require all post-secondary education and training institutions to link their datasets with tax data and to publish consistent labour market informant on employment and earnings outcomes. This is already being done as a part of a pilot by Ross Finnie with a series of colleges and universities. It should be expanded considerably. – Tyler Meredith, research director for the Institute for Research on Public Policy
More on involuntary part-time and underemployed workers
This also means more information on unpaid internships. This could tell us the average tenure in unpaid work before transitioning to paid work – and whether that's increased over time.
And data on average hourly earnings should include unpaid workers. "A rising share of unpaid work would show up in slower gains in average earnings – accurately pointing to limited bargaining power of workers and slowing costs for the employers." – Emanuella Enenajor, senior Canada and U.S. economist at Merrill Lynch in New York
Average length of time to fill a job vacancy
The enhanced job vacancy survey will ask new questions about what skills employers are looking for, and what they've done to recruit. "We need to get to a better understanding of the kind of hiring problems employers are facing, and the reasons behind them. The issue is likely much more nuanced than just whether wages or skills are sufficient to fill available vacancies." – Tyler Meredith, research director for the Institute on Public Policy
Include persons with disabilities in the labour force survey
We need better labour market data on persons with disabilities, along with information that captures the realities facing visible minorities and aboriginal populations. We can get some sense now by linking with other surveys, but we are still lacking any real-time picture necessary to see how this (and other) sub-groups are doing in the labour market. Including these groups may, however, potentially require a much larger sample than is currently collected.– Tyler Meredith
Better wage information
This would flag industries and occupations under pressure. "It is a critical ingredient for programs such as temporary foreign workers to ensure domestic wages are not being undermined." – Don Drummond
Emphasize confidence bands
Information on the standard error of the LFS "is buried deep in the report and often overlooked." But rather than stating that Canada, for example, gained 20,000 jobs, Statscan should state they estimate that somewhere between 25,500 jobs were lost or 65,500 jobs were gained (with a 90-per-cent confidence interval). "Markets put too much weight into each single monthly jobs release and the wild month-to-month swings can often mislead market participants and the public as to the actual state of the jobs market. An alternative is to report a rolling six-month moving average." – Emanuella Enenajor
Expand use of administrative data
Relying on surveys can be costly, and difficulties in getting large sample sizes mean we lack information for example, specific characteristics or areas. In addition this methodology is not really suitable to know, for example, whether temporary foreign workers have left the country at the end of their contracts. Most OECD countries provide much more detailed labour market information based on various administrative data sources. It requires co-operation between statisticians and administrations collecting the data. "But this approach can provide more precise information about small groups of people because it covers the population under regulations." – Dominique Gross, professor at Simon Fraser University's School of Public Policy.
Better data on adult education and training needs
This would include measures of workplace training, training expenditures and unmet training needs. Much of the data we have is outdated. This series of information needs to be part of a recurring, annual survey, ideally linking to other firm-level information. – Tyler Meredith
Comments have been condensed and edited.