The latest report on the health of Canada's job market landed on Thursday, with the monthly ADP Canada survey showing the country lost 10,500 non-farm payroll jobs in June.
"We saw a significant dip in job growth in Canada for the month of June," ADP Research Institute vice president Ahu Yildirmaz said in a statement.
"This decline likely reflects the impact of regulations on mortgage financing and a slowdown in consumer spending."
But is the decline real? According to Statistics Canada, the country actually added 32,000 jobs in June. ADP's report showed a notable slowdown in construction, with a loss of 5,600 jobs. StatsCan reported a whopping gain of 27,000 construction jobs in the same period, accounting for most of the month's job growth.
Watch: The most in-demand jobs in Canada in 2018 (story continues below)
The various indicators of Canadian job growth have diverged since the start of the year, and that could cloud the picture for policymakers and business investors who rely on this sort of data to make decisions.
Even within Statistics Canada's own data, the numbers are all over the map. StatsCan's two principal measures of the job market — the Labour Force Survey (LFS) and the Survey of Employment, Payroll and Hours (SEPH) — have gone in completely different directions.
According to the more timely and widely-reported LFS, Canada recorded a net loss of 28,000 jobs in the first four months of this year, while the SEPH shows Canada gained 100,000 jobs in the same period.
"The divergence is unprecedented," National Bank Financial economist Krishen Rangasamy told HuffPost Canada by phone. The two surveys have often contradicted each other before, but never to this extent, Rangasamy said.
In Rangasamy's view, the problem is that this job data is part of what policymakers and investors look at when making decisions. For instance, the Bank of Canada looks at job numbers when determining whether to raise interest rates.
The markets and policymakers watch the LFS numbers more closely than the SEPH numbers, because the LFS is more timely; SEPH numbers come out with a several-month delay. But right now, that means people are paying attention to the more pessimistic set of numbers.
Rangasamy believes the more optimistic numbers — the 100,000 job gain registered by the SEPH — are probably more accurate right now.
He points to new data this week on employment insurance (EI) showing that the number of regular EI recipients has fallen by 15.2 per cent in the past year. That suggests a strengthening job market, not a weakening one.
Earlier on HuffPost Canada:
Rangasamy also likes the SEPH better because it includes a survey of employers, which makes the data "arguably more reliable" than the more closely-watched LFS, which is based on a survey of households.
That's a point Statistics Canada seems to concede.
"SEPH is based on a census of businesses and therefore is not subject to the same degree of statistical uncertainty" as the LFS, a StatCan analyst wrote in an email to HuffPost Canada.
She offered a number of reasons why the two surveys would diverge, including that "people with multiple jobs are counted for each payroll job in SEPH, while they would be counted once in the LFS."
The analyst didn't offer any reasons why the disparity is particularly large in recent months, but noted that if you look at the long-term trend, both measures point in the same direction.
"If we look at the year-over-year change from April 2017 to April 2018, we can see that the LFS shows an increase of 220,000 employees, while SEPH reports an increase of 384,000."
A huge margin of error
Questions have been raised in the past about the reliability of StatsCan's employment data. Critics have pointed out that the Labour Force Survey completely omits First Nations territories, arguing it's too costly and difficult to collect data there.
And StatsCan's own description of the Labour Force Survey suggests that the margin of error is so large as to make monthly numbers almost meaningless. For the agency to be 68 per cent sure that Canada actually added jobs in any given month, the job growth number would have to be greater than 40,000.
But the change in the number of jobs is frequently less than that, meaning there's virtually no way to be sure whether Canada lost or gained jobs in a given month. The numbers only become meaningful when you look at a longer period of time.
Rangasamy stressed that he is not trying to discredit Statistics Canada's numbers.
"But there are question marks here," he said.
Also on HuffPost: