(Bloomberg) -- Canada’s job market is less robust than the country’s most closely watched employment indicator suggests.

Data from the payroll employment survey — a census of all paid workers in Canada — indicates the country added a mere 17,000 jobs over the six-month period ending in February. That’s a far cry from the 183,800 reported in the labor force survey, which polls households and is faster to publish results.

That difference of 166,800 jobs is more than 22 times the average gap between the two surveys since 2001, and has been growing for months.

During the height of the pandemic in June 2020, six-month job growth in the household survey exceeded payroll growth by more than 700,000. But that gap now approaches its highest level recorded before 2020, which was 186,100 in 2012.

The payroll survey relies on administrative data reported by businesses for tax purposes. It’s a direct input to the calculation of Canada’s gross domestic product.

The labor force survey, on the other hand, polls about 56,000 households monthly, allowing it to provide an earlier indicator of the pulse of the job market and capture geographical and socioeconomic differences. It posted new data for March three weeks ago showing the country lost 2,200 jobs that month.

A key difference between the two is that the payroll survey does not include self-employed Canadians, who drove February’s 40,700 job gain captured by the household survey. 

Thursday’s payroll data point in the opposite direction, recording a 17,700 job loss for that month.

In general, the two surveys don’t disagree much, Fred Demers, an investment strategist at BMO Asset Management, said in an interview.

“It gets interesting when you start to see a little bit of a gap opening up between the two, and then the safest thing you can do is bet on payrolls, in terms of giving you leading information about what’s going on.”

While the household survey’s increase in unemployment has been driven by a growing population rather than job losses, payrolls suggest those losses may soon appear in the household data, Demers said.

Self-employment and the intrinsic noise from the household survey’s small sample size may be reasons for the growing gap, he added.

Dominique Lapointe, director of macro strategy at Manulife Investment Management, agreed that the two surveys tend to track closely in the long run. 

He also agreed that the household survey has converged toward payrolls in recent years, but cautioned that month-to-month volatility in the latter has been higher since the pandemic, making it harder to tell if it’s still more reliable.

Lapointe noted that job growth in services has been faster in payrolls than the household numbers.

“What might be happening since the middle of 2023 is services sectors in the labor force survey catching up to what is reflected in the payroll survey, especially in industries such as health care, education and public administration, which continue to see rapid hiring,” Lapointe said in an email.

“What remains concerning is the lack of net private hiring since the middle of 2023 as the labor force continues to grow at a rapid pace.”

The growing divergence between strength in the public sector and weakness in the private sector is at least one area where both surveys agree. 

Read More: Public Sector Hiring Is Driving the Labor Market in Canada

While the household survey has shown a gradually rising unemployment rate and even a small surprise contraction in March, Thursday’s payroll numbers back the idea that the Bank of Canada risks keeping its policy rate too restrictive for too long, a concern highlighted by some officials in the most recent summary of deliberations.   

Both headline and core inflation are now within the Bank of Canada’s 1% to 3% tolerance band, and retail sales are on track for no growth in the first quarter. That said, GDP growth still looks strong, supported by exports and a population surge.

Policymakers next set rates on June 5. The majority of economists in a Bloomberg survey expect the bank to cut policy rates by 25 basis points at that meeting. 

©2024 Bloomberg L.P.