StatsCans job numbers a puzzle to sort out

Today is Jobs Friday, usually a pretty good day foreconomists and financial reporters alike.

The monthlyrelease of the Labour Force Survey from Statistics Canada probably getsmore attention than any otherof the agency's reports.

It sets the unemployment rate nationally and for each province and city, something that can then affect public policy, like employment insurance payments. It also tells us how many jobs were created across the country.

The payroll survey is probably the lesser of two evils.EmanuellaEnenajor, BoA Merrill Lynch

According to Friday's Labour Force Survey, Alberta lost 11,000 jobs in October, mainly in the hard-hit natural resources sector.

But over the past 10 months, something odd has been happening with the numbers in Alberta. Most months, the monthly surveyshowed the market holding up pretty well, with virtually no net job losses.

On the ground in Alberta, that seemed impossible. The province is in recession andjob cuts are regularly announced inthe oilpatch.Everyone seems to know someone who has lost a job. The energy sector estimates that 35,000 positions have been cut.

• Alberta has lost 35,000 oilpatch jobs, petroleum producers say

• Energy sector layoffs pile up as companies say efforts to cut costs paying off

There's been a lot of head scratching. Are people losing energy jobs and going to retail? Are they movingto another province? Pivoting to self-employment?

Here's where it gets interesting, though. There's the other employment report, called the SEPH, the Survey of Payroll Employment and Hours. It also comes out monthly, but to little fanfare because of the time lag, two months behind the Labour Force Survey.

However, in August, theSEPH showed that there had been 53,000 jobs lost in Alberta in the past year. The August Labour Force Survey showed that 41,000 jobs had been created over the same period.

2 reports: A difference of 94,000 jobs

That's a difference of 94,000 jobs.So what's going on? Which report is right?

The Labour Force Survey is just that a survey. Every month a battalion of interviewers contact 56,000 Canadian households to ask them who in their house has a job, along with a long, long list of other questions. Those numbers are then grossed up or extrapolated to get a picture of the whole country. Onceyou agree to do the survey once, you have to do it for six months.

• Edmonton employment centre jammed with job-seekers

• Premier Rachel Notley creates new economic development ministry

In the provinces, the sample size is smaller. In Alberta, roughly 5,700households are contacted, 0.5per cent of the population. That is clearly part of the problem.

The Survey of Employment, Payroll and Hours is different. It's based on companies remitting taxes to the Canada Revenue Agency. The CRA shares with Statistics Canadathe numbers of employees on payroll. It's a raw count of who's got a job with a company. It doesn't include household workers, the self-employed, or farmers. Itwon't tell you if the jobs were part-time or full-time, or the age, gender or education of the workers who lost or gained jobs

The first instinct is to assume the payroll report is more accurate, since all the jobs are counted. But it's not that simple.

"We doencourage people to look at both, but not necessarily to compare them because they measure different things," said Andrew Fields, an analyst with Statistics Canada, pointing out that SEPH counts payroll entries, not people.

"Someone could have two or three jobs, but in LFS, that would only be one person, in SEPH that would be two or three entries."

Which report is better?

That would likely explain some of the difference. As people lose salaried jobs, they find other work that doesn'tplace them onto a payroll. For example,Nicole Pittman, who lost her job at an ad agency and sold artisanalpopsicles at festivals through the summer.

But that probably doesn't account for a discrepancy of 94,000.

"You have a household survey that's quite volatile and a payroll survey that's lagged and subject to heavy revisions," said EmanuellaEnenajor, a senior economist withBank of AmericaMerrill Lynch Global Research.

"The payroll survey is probably the lesser of two evils."

Unless you need the unemployment rate.There is only one source for theunemployment rateand that is the Labour Force Survey, which showsAlberta's unemployment rate marching higher over the past year, from 4.6 per cent in September 2014to 6.5 per cent a year later.

The same report showsthe creation of 31,000 jobs in that period.StatisticsCanada suggests that's because Alberta's population grew over the period. So there are more people looking for work, but no new jobs created, hence the increase in the unemployment rate.

If itseems ridiculously complex, that's becauseit is, and economists generally look at both reports and over time simply hope that the circle is squared.