Why there are better ways to measure unemployment

The state of the jobs market is best assessed by a number that is not given enough attention by Statistics Canada, and the many media reports based upon its monthly press release.

The headline attention is all soaked up by the unemployment rate and the level of employment, when it really should be something Paul Krugman—the Princeton University economist and New York Times columnist—calls his “favorite gauge” of the employment situation.

The unemployment rate can be an ambiguous indicator because it depends not simply on the number of people without a job, but also whether they are looking for work. Giving up on a job hunt excludes you from the count, so the amount of slack in the labour market can be understated.

At the same time it is possible for the unemployment rate to shoot up when the economy starts producing jobs if a good many of the jobless expect their prospects to be better and begin looking for work. In this way the amount of slack in the market can also be overstated.

This is one reason why Paul Krugman focuses on something called the Employment-Population ratio, or simply the “employment rate”. Using the number of employed is misleading because it takes no account of the size of the population. The employment rate makes this correction: it is the number of employed divided by the population, in other words the fraction of the population that has a job.

Labour economists tend to look at this ratio for those in their so-called “prime” working years, usually men who are 25 to 54 years of age. This highlights business cycle changes, leaving demographic transitions associated with retirement, schooling, and, to some extent, household duties aside.

The picture for Canada shows a clear drop at the onset of the recession in late 2008, and only a tepid recovery. The jobs market has yet to return to the pre-recession level of employment when more than 86% of prime age men were working. From this perspective it appears to have stalled.

Krugman actually looks at the employment-population ratio for both men and women together. But this would change my picture only slightly: a little more continued growth since January 2011, but still no return to pre-recession levels.

The point is that the employment rate leaves us with a very different impression than the lead visual in Statistics Canada’s press release, which is simply the total number of employed. Here it is:

Source: Screen Shot from Statistics Canada, The Daily, September 7th 2012. The last data point is August 2012.

The media are quick to reprint this graphic, as the hard-copy edition of The Globe and Mail did this Saturday in a story that, ironically, was focused on the stalled recovery.

This offers no correction for the size of the underlying population. It shows a complete recovery of the number of jobs lost during the recession by late 2010, and that is certainly one reason why it is a popular talking point for many politicians.


11 thoughts on “Why there are better ways to measure unemployment

  1. Dear Michael: I find that the UR and the ER are pretty much mirror images of each other and that neither, on their own, are great portraits of our labour force situation. The 20/20 files on LFS in the Territories that I get from StatCan also allow me to break out Abo/Non-Abo, by age cohort and by gender. However they also give me indicators such as Participation Rate as well as Employment Rates, which I find to be very useful for looking at the engagement of sectors of the labour force in the economy. If you look at LFS publications by the NWT Bureau of Statistics they do use ER rather than UR.

    The Yukon Bureau of Statistics provides all of these breakdowns for the Yukon through its highly accessible website: http://www.eco.gov.yk.ca/stats/stats_employment.html

    For example, breaking out Aboriginal PR male/female starts to clearly illustrate the failure of our government programs on many fronts to integrate Aboriginal males into the labour force compared to females, The PR rates for men are very low, aboriginal women here appear to be providing the stable labour force in the First Nations communities, regardless of the their extra duties in the reproduction of labour.

    Your Arctic Correspondent
    Greg Finnegan, Ph.D.

    1. thanks for this, and for the useful link. I have often wondered how applicable the Labour Force Survey concepts are in the Yukon, NWT, and Nunavut. Any thoughts on this?

      1. Hi Miles: Everything you wanted to know about Territorial stats and more is attached. LFS is run in the North but it is a different beast due to sample size.

        It reports separately form the National. Do you use the StatCan 20/20 files?

        We can discuss over a coffee sometime soon, as I may be heading to Ottawa with a client, want a free lecture?

        This is work that I did for CanNor in an attempt to get them to hire a really hard working Arctic Research guy in his mid 50’s! to run their non-existent research shop, how can you be a Regional Economic Develop agency and do no research?

        It needs about 30K to update it and make it into a ppt presentation, then maybe 60K more to make it into a Hans Rosling TED Talk style presentation:

        http://www.youtube.com/watch?v=3qRtDnsnSwk &feature=relmfu


        Greg Finnegan, Ph.D.


        Changing Climate Consulting

        10 Rosewood Place

        Whitehorse, Yukon, Y1A 4X3



  2. Keep pushing on this, Miles! The ER (for ‘working age’ or better ‘prime working age’, not for the population as a whole) ought to be the consistent lead in the STC monthly reports, perhaps together with UR. And the ER ought to be provided up close to the front for two or three pop groups even if ‘prime age’ ER should be the headline. Same for international comparisons. In fact, the STC monthlies are dismally outdated in their data benchmarks, especially the headline data (number of jobs, of all things!). And it is definitely not good enough to provide the ER data for experts, buried somewhere in the LFS: this is not a research issue but an issue of public understanding (including the media), public debate and public policy.

    Good for you!

    BTW, get on their case also re student debt numbers: even STC regular releases promote by selective ‘headlines’ the impression that ‘average student debt’ of, say, $27,000 means debt of the ‘typical’ graduate, which is of course nonsense since — you know better than I — about half of them graduate with no debt, the ‘average’ is only for the half with any debt, and that ‘average’ is much higher even than the more ‘typical’ mean among those with debt. Not to say student debt is unimportant, but let’s try to focus on facts not factoids. (And this should not be seen as a Quebec-compared-to-ROC issue, an impression left by debate earlier this year.) Or am I outdated on this one?

    Andrew Johnson

What do you think?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.