Statistics Canada's latest data on Canadian employment suggests that the country's hiring may have been dramatically weaker over the summer than previous numbers had indicated, raising questions about just how well Canada's labour market has recovered from the economy's second-quarter slump.
Statscan's monthly Survey of Employment, Payrolls and Hours (SEPH), which uses employer data to gauge employment levels on private- and public-sector payrolls, reported that the country's payroll jobs were down 50,400 in August compared with July.
The report was wildly at odds with Statscan's earlier and more closely followed Labour Force Survey (LFS) for August, released in early September. The LFS, which polls Canadian households, estimated that the country's employers (excluding the self-employed) increased their job count by 65,400 in the month.
That raises questions about just how healthy the economy was this summer.
Most experts believe the summer was a period of strong growth as the country rebounded from this past spring's devastating Alberta wildfires that dragged the economy into a second-quarter contraction. The August LFS findings had put a stamp of approval on the summer-recovery theme; but the SEPH data raises the possibility that the labour market might not have bounced back at all.
"The question is, which is correct? There's no easy answer," said Krishen Rangasamy, senior economist at National Bank of Canada in Montreal.
It's not unusual for the employment figures in the two surveys to differ in any given month, as they rely on different methodologies, different sources of information and different timing of their data collection. But the size of the discrepancy between the SEPH and the Labour Force Survey readings in August even raised eyebrows within Statscan, said analyst Gordon Song of Statscan's Labour Statistics Division.
"It was discussed at length," Mr. Song said. "It is something we are looking at."
Markets tend to give more weight to the LFS, as it is the much more timely labour indicator of the two – typically released within a week or so of the end of each month, compared with a lag of nearly two months before the SEPH is published. But the LFS, a survey of about 56,000 households, contains a large statistical margin of error: The August figure of a 50,000 decline in employment on payrolls, for example, was considered accurate to within plus or minus 37,600, about two-thirds of the time. As a result of the potential for statistical and sampling anomalies, the LFS has developed a reputation for month-to-month volatility and unpredictability.
Many economists put more faith in the SEPH, which uses actual company payroll tax filings in addition to surveying employers on the actual size of their payrolls, and thus is considered by many to be the more rigorous measure of the two. But the SEPH is subject to sometimes sizable monthly revisions, as additional tax-filing information comes in to further flesh out the payroll picture. For example, the SEPH's change in employment in July, originally reported as a loss of 1,600, was revised in the latest release to a drop of 19,300.
Mr. Song suggested that sampling error in the LFS, or incomplete tax data used in the SEPH, or both, could have contributed to the wide August discrepancy.
"As we get more data, [the SEPH] could be more in line with the LFS," he said.
Over time, the trends in the two measures have historically tended to converge. But in the first eight months of 2016, where the SEPH has shown average monthly payroll job growth of a paltry 3,700, the LFS has produced a much healthier monthly average of 14,800.
Mr. Rangasamy believes the LFS may be overstating the strength of Canada's labour market, noting that the vast bulk of the job creation it has reported for the year to date has been part-time work.
"Just look at the [economic] growth," he said, noting that other indicators have pointed to a relatively sluggish economy with which slower job growth would be more consistent.
"The real answer is probably somewhere between the two," he said.
The SEPH report comes just days before Statscan is scheduled to release August gross domestic product numbers, next Tuesday. While the employment number raises the possibility that the month was much slower than believed for job creation, the SEPH also reported that average weekly earnings surged 0.8 per cent from July, the biggest one-month gain in nearly a year – a welcome development for consumers in the month.