Precarious work, Federal government edition
There was a recent article in the Hill Times about temporary workers in the federal public service, noting that this number is growing even under Trudeau’s sunny ways (that’s not entirely fair, the report only covered the first 5 months of the Liberal’s tenure).
The numbers come from the Privy Council clerk’s annual report, which shows that the number of temporary and contract workers in the federal public service increased by 2,800 between March 2015 and March 2016, to 35,000 workers, or about 13% of the total federal public service.
Because the recent Changing Workplaces Review from Ontario was on my mind, and the recent attention on the abuses of temporary employment agencies, I wondered if we even knew how many temporary agency workers there are in the federal public service, or the federally regulated sector.
From the Annual Survey of Service Industries: Employment Services, we do know the size of the Employment Services Industry – $13.3 billion in 2015, and we know that over half of that was temporary staffing. Government and non-profits make up about 10% of sales – but there would also be temporary agency workers placed in transportation and telecommunications, for example, and that breakdown wasn’t available here.
I had better luck with the Federal Jurisdiction Workplace Survey (FJWS). This survey covers industrial companies under the federal jurisdiction (not workers directly employed in the federal service). In 2015 the FJWS asked about the number of temporary workers paid through an employment or personnel agency over the course of 2015. Employers reported 60,000 workers paid through temporary agencies, most of these workers were employed with large workplaces. The breakdown by company size and industry is shown below.
Distribution of temporary workers paid through an employment agency, 2015
| # of employees | ||
| Company size | ||
| 1 to 5 employees | 900 | |
| 6 to 19 employees | 800 | |
| 20 to 99 employees | 1,300 | |
| 100 employees and more | 57,100 | |
| Industry | ||
| Air transport | 800 | |
| Rail transport | x | |
| Road transport | 4,600 | |
| Maritime transport | x | |
| Postal & pipelines | 39,700 | |
| Banks | 8,100 | |
| Feed, flour, seed & grain | 300 | |
| Telecomm & broadcasting | 6,000 | |
| Miscellaneous industries | 300 | |
| Total | 60,000 | |
Source: FJWS (2015)
Note: The “x” indicates that the data has been suppressed due to confidentiality concerns as required by the Statistics Act.
This all tells me two things. First, it is totally reasonable to look at ways to protect precarious workers in the federal public service and the federally regulated sector (remember all that noise about the NDP federal minimum wage promise not helping anyone?), and second – we need better data to do it.
A lot of the temporary workers my be making well above minimum wage, although many probably still are precarious workers.
For a while my wife was a “temp” working full time at a serious office job for well above minimum wage, but they still did illegal things like not provide holiday pay.
Do you think those Postal & pipelines are more postal or more pipelines?
We do need much better numbers in this area. There is not a lot- and if we are going to monitor low wage workers and minimum wages – it is critical we expand the measurement of such. Statistics Canada last redesigned the LFS in 1997- and that was just to bring in a hand full of changes. Now with precarious work becoming such a larger issue in society- we need to bring the LFS and other measurement tool to bare on this issue. I would recommend to anybody on this list it you sit on the Advisory committee for labour Statistics at Statistics Canada- that you try and work at such efforts. Without better data on this issue we cannot direct policy with the focus that it needs.
For example- yes the monthly survey does measure wages- but the experience at Statistics Canada is that the wage variable has a lot of non-response. Also- the wage question is not asked for those respondents indicating they are self employed. I estimate with the LFS micro file that in April 2017 nearly 3.5 million employees- make less than $15 per hour. I ran it for every month in 2016 and yes there is a seasonal component to it and will be publishing something on this soon. The problem is- we have around 3 million self employed that do not have a wage variable. We know from many studies- that self employed make less than employed on average- and also contain large pockets of the precarious workers- mainly identified in the LFS as unicorporated and without employees- which number around 1.3 million.
I am using a neural network right now to train on the micro file of the LFS at predicting low wage workers- so one can try and model an estimate for these workers. So far using a Tensor flow and a multilayer perceptron- with backpropogation- with one hidden layer- using a sigmoid functions on 55,000 workers- I am estimating at a 85% success rate on predicting low wage workers. Interesting as I am only using SEX, AGE Group, Usual hours worked, Tenure, and Union Status. I will be adding in an industry NAICS one-hot encoded – which I am sure will boost the model further.
But we should not have to use such AI based models to predict such! We should have such data. I do think one could get it from the Tax data- but they rarely ever share such data- and most times it is 3-4 years old by the time you get it.
oh and I will be adding the new immigration variable into this model which I am sure will help the classification rate for the machine learning. Statistcis Canada has redone the Micro file for 2017 and has included the immigration status in the 2017 micro file- as well as the new occupation codes that were pulled from 2016. I will be publishing the results on my website in the near future. If anybody wants me to dig out some data on such research- let me know- if it is not too much work- I can do it for you.