I noted a couple of months ago that employment losses in the recession were generally concentrated in the goods-producing sector: some 350,000 jobs were lost there, and there's been little in the way of recovery. On the other hand, job losses in the services sector (about 3 times as large) were on the order of 100,000, and they've all come back: employment there is now higher than before the recession hit.
But here's the thing: as of the February 2010 release, the increase in unemployment in goods-producing sector is only one-third that of the numbers of unemployed service sector workers. What's going on?
Here is the change in employment in the two sectors since the October 2008 employment peak:
And for the number of unemployed by sector (all data taken from CANSIM Table 282-0094):
The story in the services sector is simple enough to tell, I think. Net job creation is positive, but not strong enough to absorb the usual flow of workers into the labour force.
Let's break the goods sector down a bit. Here is employment within the goods sector:
The manufacturing sector accounted for just over 10% of employment before the recession, and it absorbed almost two-thirds of the job losses.
And here's the graph for unemployment:
Even though there are 200,000 fewer jobs in the manufacturing sector, the number of unemployed manufacturing workers is back down to pre-recession levels. How did that happen?
- One answer might be that they are looking for work in other sectors, so are no longer classified as being unemployed goods-producing workers. But from what I can tell from the LFS questionnaire, the only industry-specific information asked of those who are unemployed is where they last worked, not what sort of work they are looking for.
- Another answer might be that they have simply dropped out of the labour force. But you'd expect them to exhaust their EI benefits before making that step, and I don't think (numbers, anyone?) that there were that many people who started to exhaust their benefits in mid-2009. If the decline had started (say) six months later, then I think this would be the most plausible explanation.
- One thing that might be going on is that older manufacturing workers are retiring (with varying levels of persuasion and/or enthusiasm) and are being replaced by younger, unemployed workers.
But I don't really know. And Statistics Canada doesn't have the budget to track this sort of data - although certainly Human Resources Canada could provide data on the number of people whose EI benefits were being exhausted.
Anyone got a better story?
Two thoughts:
- In the vein of no longer classified as goods-producing, is it possible that a number of people who lost long-term manufacturing jobs later found short-term service jobs, and when their contracts ended because classified as service workers?
- Any chance that there are different surveys with different definitions of service sector? I would consider everyone at my engineering and surveying company to be service workers, but if construction is part of goods production, then some portion of our work may be classified differently. Same thing at drilling companies, as drilling companies are generally considered "oilfield services," but they are actually constructing wells, so a tangible good is produced. There's probably some other industries like that that I'm forgetting.
Posted by: Neil | March 15, 2010 at 02:47 PM
This is highly informative material. With almost two-thirds of the jobs lost among the 10 percent of the labour force in the manufacturing sector, that suggests Southern Ontario has taken most of the hit, and, of course, it has the largest proportional population size, so the unemployed become less visible.
When Newfoundland lost the cod fishery, it was said at the time the impact was as if Ontario had lost auto. The cod have not come back, what are the prospects for auto (and steel) ?
Posted by: duncan cameron | March 15, 2010 at 03:38 PM
On the subject of the StatsCan budget - where was the urgent need for these new economic data statisticians over the past 10 years? How did StatsCan neglect to hire them during the largest increase in the public service in a generation?
I guess all their new hires were in the "special pleadings" division.
Posted by: Geoff NoNick | March 16, 2010 at 09:10 AM
I recall last summer that the Labour Force Survey recorded a significant shift into self employment among men aged 55+. At the time, I assumed these people were financial services or professional/sci/tech workers. But maybe many were former manufacturing types?
Also, I've only been watching LFS numbers through the NAICS codes based view, but maybe the occupation (NOCs) code perspective would shed some light on your question?
Posted by: Wendy | March 16, 2010 at 12:24 PM
That's a thought. Ideally, we'd want both the NAICS and NOC, but apparently you have to pay for that.
I'll take a quick look. Thanks.
Posted by: Stephen Gordon | March 16, 2010 at 12:35 PM
Nope. At the monthly level, they only have 10 or so occupations, such as "Occupations unique to processing, manufacturing and utilities". At the annual level they break it down more, but it's harder to extract the business cycle out of that. In any case, the 2009 numbers aren't out yet. I'll try to remember to get back to this.
Posted by: Stephen Gordon | March 16, 2010 at 12:46 PM
Another idea (related to previous suggestion): the Labour Force Survey records what industry people were working in, not what they did. Is it possible that among those laid off in "manufacturing" were accountants or bookkeepers, or supply clerks, or administrative? That is, many laid off manufacturing workers had skill sets that were more universal and thus were able to shift into another industry type.
I have some older NOC vs NAIC numbers -- I'll take a look and see if there is anything to learn.
Posted by: Wendy | March 18, 2010 at 03:45 PM
I just looked at the manufacturing related series for both, for the Toronto CMA (what I have) to April 2009 (last time I downloaded NOCs). From 2004 to 2009, the NAICS code employment levels went from 500,000 to 360,000 meanwhile the NOCs code "occupations unique to manufacturing etc." went from roughly 275,000 to 150,000. If my math is right, this suggests that 15,000 of those laid off "manufacturing workers" actually did something not "unique to manufacturing" which may mean they found employment elsewhere.
I think there are still a lot of "missing" manufacturing workers.
Posted by: Wendy | March 18, 2010 at 03:59 PM