OTTAWA - A cloud of doubt settled over Canada's dismal July jobs numbers Tuesday as Statistics Canada announced that it made an error in formulating the figures.
Corrected data will be released Friday, one week after the agency reported a measly 200 positions were created in July — a number that fell woefully short of expectations and left the Conservative government reeling.
"We had an independent analyst that was working on something else and by comparing numbers, he signalled to us something after the release of the data," Sylvie Michaud, director general at the education, labour and income branch of Statistics Canada, said in an interview.
"So we investigated and we've been able to isolate the error. We know where it happened in the process, so we're rerunning all the tables to prepare for another release this Friday."
Michaud offered no hints about whether the corrected labour force survey would amount to good news or bad news for the Canadian economy. The monthly jobs numbers are a significant economic indicator.
The news of a formulation mistake came as Finance Minister Joe Oliver met with guests at a two-day summer retreat in Wakefield, Que., to discuss the Canadian economy. Job creation is one of several key items on the agenda.
Economists had expected the economy to bounce back from an unanticipated decline of 9,400 jobs in June, and add as many as 20,000 new jobs the following month.
Instead, Statcan reported that the number of full-time jobs fell by 59,700 while part-time jobs increased by 60,000 — figures it now suggests were faulty.
Soon after Statistics Canada released its July figures last week, Scotiabank economists suggested they didn't add up.
"What caught our attention was a series of remarkable coincidences that — while not statistically impossible to explain collectively — nevertheless arise with the regularity of asteroids hitting our planet," economists Derek Holt and Dov Zigler wrote on their Scotia Flash report.
Among the peculiarities, they noted, was the fact that full-time job losses were almost exactly offset by part-time job gains, while the drop in goods-producing jobs was precisely matched by a rise in service jobs.
As well, they pointed out, "35,400 people left the workforce all because they voluntarily counted themselves out of searching for work as the number of unemployed dropped by 35,700 despite no job growth during the month."
The Statistics Canada report also suggested 42,000 construction jobs were lost in July, even though housing starts rose for the month.
"It's pretty rare — so rare that I can't recall a prior occasion," Holt said of the error.
In an updated report Tuesday, Holt also questioned Statistics Canada's decision to wait until Friday to release the corrected figures when "they know they have an error now."
The mistake, once unheard of for an agency that used to be heralded as among the best of its kind in the world, is just the latest in a series of quality-related controversies that have dogged Statistics Canada in recent years.
Earlier this year, auditor general Michael Ferguson concluded that the agency's top survey on job vacancies is too vague and provides little value to governments and other users.
Ferguson said the Statistics Canada survey of employment, payrolls and hours doesn't provide specifics on the precise location of job vacancies within a province, and answers few questions about which particular jobs need filling.
Another Statcan survey commissioned by Employment and Social Development Canada, which included input from 25,000 employers on workplace demographics and skills requirements, never got past the data-collection phase because funding ran out, Ferguson noted.
The auditor recommended the agency determine whether to keep certain surveys alive even after a department or organization stops providing funding. Statcan has been grappling with a $29.3-million funding cut over the last two years.
The Conservative government's decision to do away with the mandatory long-form census questionnaire in 2010 cast doubts on the quality of the data produced by its replacement, the voluntary National Household Survey, in 2011.
The following year, during the release of data from the 2011 census, Statistics Canada was forced to delay for a month the release of critical NHS data when agency experts found mistakes during last-minute checks.
Follow Lee-Anne Goodman on Twitter at @leeanne25