What’s Happening in the Job Market?

There’s a bit of a mystery in the stats Erin and I have been puzzling over.

According to the Labour Force Survey,  the number of employees in Canada fell by 79,000 in July. According to the Survey of Employment, Earnings and Hours, the number of employees rose by 74,000 in the same  month.  The methodology of the surveys is quite distinct – the former is a household survey and the latter a survey of employers. But usually, especially of late, they have moved more or less in tandem so a difference this large is puzzling.

Other sources of information for the month don’t help much.  GDP stagnated in July which suggests little new hiring, but the number of new EI claims fell.

No doubt it will all average out in the end, and the Summer will come to be seen as an end to huge layoffs at least for now. That is suggested by the fact that the LFS registered a modest gain of 38,000 employees in August after the sharp fall in July.

Nonetheless, I still expect we will hit a double digit unemployment rate over the next few months.


  • It would be nice if we could get some timely data on welfare claimants so that we could make some conjecture about how useful the EI claims series is at this stage.

  • Seph is mainly a payroll based using admin data for the majority of it survey methods- a small sample survey is completed to fill in the missing benefits.

    As you stated there is the occasion the two become out of sync- which happened a few years back during an election at the beginning of the last bubble. Seph was showing big increases and LFS was flatlined. I recall everybody be up in arms at it.

    My guess would be SEPH would be faster to pick up on changes than LFS due to the LFS having the rotational based sample- i.e. only a select portion of the sample is new each month. THis as I stated previously smooths out the month to month, but at a cost, i.e. you may be smooth out the variance a bit but your estimate could be a lot further away from the true population amount, given the alternative of drawing an entire new sample each month. However this technique has its uses.

    My hunch is to trust SEPH-admin data is always more accurate if it is properly clean that is- but recall, a payroll job is not the same as a an employed person in the LFS. THey basically take the increased payroll total and through an algorithm estimate the payroll jobs.

    Potentially we are seeing a rise in payroll without the hiring, however I cannot see it being that much.

    It could also just be either a non-sampling error like a processing error or it just may be plain old sampling error in LFS. But again I woudl not think the error could account for that much of a swing. Must be a combination of much of the above.

    I am surprised they have not addressed the difference given the highly charged political climate.

  • oh and Travis, I think there is a way they could obtain the information on benefits expiring versus finding a job from the EI numbers. It would not be clean but there is an estimate that could be had. They should have some kind of end of benefits question put in place.

    I am sure each province has a data set on social assistance counts. I know for sure at the municipal level the counts exist- cause they pay the bills. I’ll today and have a look.


  • Hi Travis and anybody else.

    Ontario has a monthly tabulation of its Ontario Works programs- which I believe is different from social assistance.

    I will keep looking.

    Anybody else have info in social assistance numbers?


  • sorry travis, Ontario Works is a program that is coordinated and funded in cooperation of the provinces and municipalities where possible. In rural areas it is the province. So I guess that doc are the numbers we were after. As you can see there is a bit of a spike starting in the fall so it looks like those that did not have EI must have went directly to social assistance, and it also looks as though it has been increasing as the recession has worn on.

    Somebody must collect them at the national level.

    I will look around at statcan.

  • What I cannot figure out is why the provinces have not been more vocal on expanding eligibility and duration for EI given the EI program is a federal program and welfare is provincial. From a provincial point of view the more restrictive the EI program the greater the provincial welfare bill.

  • There has typically been a lag between onset of recession and increased use of social assistance since EI exhaustees must exhaust their financial assets first (and since many EI exhaustees are ineligible for social assistance because there is another earner in the household.)

  • But that is exactly my point! The provinces have an interest in extending both intake and duration of program. If, as is the consensus, the job market is going to be lousy for another year but that the economy is picking up the Provinces would be smart to be pro actively agitating for the expansion of EI so that eight months from now they are not on the hook. I guess what I am trying to say is that the provinces could attempt to offload their downside risks.

  • Hey Travis,

    Did somebody on this blog prod the media lately- look at this article on exactly what we were discussing above- i.e. the lack of stats on EI.

    Nice to see a reporter pushing the statistical space with their pens once in awhile. Makes left stats guys like me smile- yah I know odd things like that actually make me happy- go figure huh.

Leave a Reply

Your email address will not be published. Required fields are marked *