Updated: Thu Nov. 05 2009 5:22:27 PM

The Canadian Press

TORONTO — The surprisingly sharp turn-around in Canada's labour market this summer is raising eyebrows among analysts that the most timely and critical barometer of economic strength may be sending false signals.

A consensus of economists surveyed predicts that Friday's new labour force survey of employment will find that 10,000 new net jobs were created in October, the third straight month of gains.

The likely result from such a finding is that the Canadian dollar and stock prices will strengthen moderately in Friday trading, since job creation is the most important indicator of the health of an economy.

The thinking is more jobs means more goods being produced, higher company profits, and more money in Canadians' pockets for buying televisions, cars and homes.

But Scotiabank senior economist Derek Holt, among others, is leery of the Statistics Canada labour force survey that has shown 58,000 new jobs created in August and September, when according to the same government agency, Canada's economy was flat in July and contracted by 0.1 per cent in August.

"Why would employers be adding so much to their payroll when the economy on balance still stinks?" Holt asks.

"Don't tell exporters that the economy is wonderful. And the natural resources sector, they're not going to get to sustained hiring on modest increase in commodity prices. Intuition would suggest there is cause for deeply discounting what the labour force survey is saying."

The recent employment gains are also at odds with the general belief that as the economy recovers, jobs are one of the last indicators to return since firms will resort to increasing the hours of the employees they have before adding workers.

Of course, there is more than intuition casting doubt on the rosy data.

Although it gets little attention, Statistics Canada puts out a second barometer of employment every month called the Survey of Employment, Payrolls and Hours which counts jobs from the industry perspective, rather than the better known survey of households.

The industry survey suffers from the disadvantage of coming out a month later than the household poll, hence is regarded as stale news and does not measure the variable self-employment category.

But that may be what makes it a better indicator of the economy during this current volatile and uncertain period of halting recovery, says Holt.

Significantly, the industry survey has been wildly at odds with the widely-reported labour force numbers this summer, and especially in August.

In the month when Canadians were told 27,000 new jobs were added, the first reversal in almost a year, the industry survey was showing that employment had fallen by a staggering 110,200 jobs.

Statistics Canada explained that half of the losses stemmed from a drop in employment in educational services, which saw an unusual rise the previous month. But economists point out that even after accounting for the anomaly, as well as self-employment, the industry survey still shows a loss of more than 50,000 jobs, not a 27,000 gain.

"Whenever you get turning points in the economy, it is quite conceivable you get these pretty hefty discrepancies, but still they are showing a completely different situation," notes Michael Gregory, a senior economist with BMO Capital Markets.

"It is a bit of a mystery and I've been warning my (colleagues) there is a downside risk here. GDP has surprised on the weak side and it is quite conceivable the job picture is weaker also."

A survey of manufacturers released Thursday also suggests things may not be all that rosy on the factory floor, given continuing weak demand for Canadian goods in the U.S. and a stubbornly high dollar that prices their products out of foreign markets.

The sector has lost about 200,000 jobs in the past year, and about one quarter of the over 700 companies surveyed in early November say they plan to cut more jobs over the next three months.

So which survey is right? CIBC economist Benjamin Tal refuses to take sides, saying Canadians should be wary of monthly numbers from both surveys and look primarily at trends in the data. He points out the error factor in the household survey is plus or minus about 30,000 jobs, meaning Friday's published employment number could be wrong by up to 60,000.

That doesn't mean markets won't react to Friday's data, however.

"Markets have to trade on something," he says, even if investors are aware the hard numbers hide built-in softness.

One of the most notoriously unreliable report is the initial U.S. gross domestic product data, which is often significantly changed the following month.

"The markets trade on the early number and they trade on the revised number," said Tal.