PERPLEXED BY CANADA'S JOBS DATA?

By Don Curren

Canada’s key jobs report – the monthly labor force survey – is notoriously volatile, and some economists question its reliability.

We take a look at what’s caused some recent consternation over the data — and how economy watchers can deal with the perplexing report.

The confusion has been intense over the last while, as the jobs numbers have flip-flopped on a monthly basis between mammoth gains and elephantine losses.

The last report showed Canada lost 54,500 jobs in March after gaining 50,700 new jobs in February. That came after a loss of 22,000 jobs in January, which was in turn preceded by some substantial gains.

Economists have been skeptical at each stage.

That skepticism is based on the methods Statistics Canada uses to arrive at its employment numbers. The LFS, also called the “household” survey, is based on a poll of a relatively small number of households — around 56,000.

There’s no guarantee the sample will accurately capture employment trends on a monthly basis, as StatsCan acknowledges.

“People don’t appreciate that in all economic statistics–but especially the labor force survey–there is a great deal of noise,” said Philip Cross, research coordinator at the Macdonald-Laurier Institute, an Ottawa-based think tank.

Some of that noise originates from the fact the survey covers only a one-week period for each month, Mr. Cross, who was formerly chief economic analyst at StatsCan, told Canada Real Time.

In an economy where there’s increasing activity in resource and construction, there’s more and more variability in how much work is actually going on in a given week, he said.

“These guys work outdoors, and they’re subject to weather,” he said.

Even Alberta’s oil-sands plants are subject to variability because of regular maintenance shutdowns. If the survey happens to fall on a week with a lot of shutdowns or holidays, the validity of the data would be compromised, Mr. Cross said.

StatsCan’s other, lower-profile, jobs data — the “establishment survey” — tends to be more stable. It’s based on a survey of business, like the nonfarm payrolls report in the U.S.

Some economists have suggested releasing the two on the same day would help, but Mr. Cross said it will likely never happen.

But instead of asking financially challenged StatsCan to do things differently, Mr. Cross suggests the onus is on those consuming the data.

Instead of reacting to one report in isolation — something traders do frequently, judging by market volatility after the data’s released — he recommends looking at them in the context of the survey results for the preceding months, and of other data.

“You risking a very big mistake if you let one number — or even two — alter your view,” he said. “I don’t hang on every single number. I think of the economy as a very fundamental, underlying trend.”