Job tenures and the gig economy

A few weeks ago, Alex Usher drew my attention to this post by the Pew Research Center, on job tenure patterns of 18-35 year-olds in the United States. The takeaway point was that, contrary to an oft-repeated narrative about the "new gig economy", job tenure patterns among millennials resemble those of the generation previous.

Of course, Canada is not the United States: what do job tenure data look like for younger cohorts up here? It turns out that this may be one of those rare cases where Canadian data are richer than American data: the Labour Force Survey has been asking about job tenure since 1976, and we can focus on more tightly-defined age groups.

I've calculated three job-tenure measures:

• Median job tenure
• Proportion with job tenure of one year or less
• Proportion with job tenure of five years or more

These are broken down by age group (25-29 and 30-35), by sex, and by educational attainment (high-school graduates and university graduates). I've also limited attention to full-time workers, and excluded the self-employed. I've put everything in an excel file you can download here. The excel file also includes average job tenures, but I won't be talking about them here.

I'll start with 25-29 year olds. Here are median job tenures for both sexes, broken down by educational attainment:

Since university graduates enter the labour force later than do high-school graduates, it's not surprising that high-school graduates in their late 20s typically have longer job tenures. I don't see a secular trend in median job tenures among university graduates, but there looks to be a slight decline among high-school graduates.

But I wouldn't be quick to conclude that high-school graduates in their late 20s in were better able to find steady jobs in 2016 than 20 or 30 years ago: those increases in median job tenures in the mid-80s and mid-90s occurred during recessions and record-high youth unemployment rates. I suspect that what was really going on was that young workers were clinging to jobs that they didn't particularly like, because the prospect of finding a better one was so bleak.

Let's break these down by sex. High-school graduates first.

There's not a lot of difference here between men and women. Here are the proportions with job tenures of less than one year:

This seems to be telling the same story. Again, the lower numbers in the 80s and 90s probably reflect weak job creation, and not stable employment: fewer workers had short job tenures, because there weren't that many new jobs.

And here are the proportions of people with job tenures of five years or more:

Once again, those increases in the 80s and 90s correspond to weak labour markets. I don't see good, stable jobs there; I see people clinging to whatever work they had.

On to university graduates aged 25-29. Here are median job tenures

I don't see a secular trend there; all I see is that counter-cyclical pattern in job tenures.

The proportion of university graduates aged 25-29 with job tenures of less than one year:

This ... looks like noise around a constant. No trend, and it's hard to pick out the business cycle.

The proportion of university graduates aged 25-29 with job tenures of five or more years:

Unsurprisingly, these rates are less than half of those for those with a high-school diploma, and again, you see that counter-cyclical pattern of spikes when labour markets are soft.

Okay, so much for those aged 25-29. I don't see much of a downward secular trend in job tenure measures for university graduates, and much of the downward trend among high-school graduates is probably explained by the supply side of the market, not the demand side. Younger workers in the 80s and 90s had longer job tenures, but that may have simply reflected an inability find better jobs than the ones they already had.

On to 30-34 year olds. Here are median job tenures:

All I can see here are those counter-cyclical swings: median tenures increasing during recessions and declining during recoveries. I'm really having a hard time with the idea that longer median job tenures are a sign of a healthy labour market. And there's not much to choose between current median tenures and those observed 40 years ago.

Again, high-school graduates first. Here are median job tenures, broken down by sex:

What's interesting here is the gap between men and women. Median job tenures for men and women high-school graduates in their late 20s look almost the same throughout the sample. But when they reach their early 30s, there's a huge gap in the first part of the sample - and then convergence. My first-pass conjecture is that this is a story about trends in child-bearing and child care/home production, but I don't feel confident enough about the topic to explore this point any further.

Here you do see a more pronounced secular decline for men's job tenures, albeit complicated by the counter-cyclical pattern noted earlier.

The proportion of high-school graduates aged 30-34 with job tenures less than one year:

I don't see much of a trend for women, and the lower rates among men in the earlier part of the sample could be a reflection of slow job creation.

Proportion of high-school graduates with job tenures of five years or more:

This basically looks like the chart for median job tenures.

And now median job tenures for university graduates:

The counter-cyclical pattern is particularly striking here. And as was the case for high-school graduates, you can see a secular decline for men, but not for women.

The proportion of university graduates aged 30-34 with job tenures of one year or less:

No secular trend jumps out at me here.

Proportion of university graduates in their early 30s with job tenures of 5 years or more

Again, to the extent that there is a secular decline, it's confined to men.

And that's it.

Here's what I get out of all this:

1. It's not clear to me that longer job tenures are a necessarily a sign of a strong labour market generating stable jobs. The counter-cyclical patterns suggest that longer job tenures may be a reflection of an inability or unwillingness to find a better job.
2. You can see some evidence of a a secular trend towards shorter job tenures, but only for men.
3. I don't see how you can tell a compelling story of a fundamental transformation towards a 'gig' economy in which more and more people are working at a series of jobs at the beginning of their careers using job tenure data.

 

 

Comments

[Jane]

"I've also limited attention to full-time workers"

You do realize that a gig economy has to do with precarious PART TIME work, right? So whatever your data indicates, it is unlikely to have much to do with reality for the workers actually employed in the gig economy.

[Bob Smith]

Jane,

The "Gig" economy isn't limited to part time workers. It would pick up people working full-time on short-term contracts (I.e.., moving from gig to gig). If the phenomenon is material it should show up in full time employment stats.

[Bob Smith]

The bigger problem is likely to be the exclusion of the self-employed - since they are likely to be employed in the gig economy as independent contractors. They are almost the definition of the "gig economy" worker.

[Stephen Gordon]

That's sort of why I added the qualifier "with job tenure data". To extent that gig=short-term employment contract, these data should pick that up. But I don't see how you could extract information about a gig economy from self-employment data.

[rsj]

This is a good point about the identification problems of reduced labor market dynamism versus gig economy type jobs.

But I wouldn't look at tenure. Kinda the point of the gig economy is that you can't get fired if you are an independent contractor, so why would you look at tenure to begin with? The fears of the gig economy are about insecurity of *income*. The Uber driver has a guaranteed job, but not guaranteed income.

Unfortunately I'm not sure how you would measure it. Aggregate income is going to be more stable than individual income. That's the point -- a firm can hire many people and pay them each a stable, lower, income than what they would each make as independent contractors, and it makes sense for the firm to do that because in a fair bargaining situation it would need to compensate the contractor for the lack of a guaranteed income.

But when people are really desperate, they will take the lack of guaranteed income over no income, so when firms have a lot of bargaining power, they are able to hire independent contractors without paying them a premium, in which case it is a no brainer to push the income risk onto the employee if you can.

I don't think there are many labor markets where firms have this much power, but it is widely believed to be happening more now at the bottom of the income distribution. IMO, this is what people mean when they say stable jobs are disappearing, and it has nothing to do with tenure.

[cm]

Also "self-employed" or "consultant" are also often just euphemisms for "not employed" (a different concept from "unemployed", but there is overlap).

I agree with others that gig work is strongly correlated with not being "properly" employed.

[cm]

Stephen: With much employment being "at will", more reliance on contingent workers (whether hired direct as "independents" or through a staffing agency), companies not investing much in their workers, and the main method to get career advancement/development and pay raises being leaving for another company, the meaning of "gig" has been broadened to refer to jobs in general.

At the same time, all this is more narrative than reality - e.g. many more people "know" that you have to switch jobs than are actually doing it.

I have some anecdotal evidence with people who grew up in this low-certainty, stagnating job environment, who generally consider their current/next job as a stepping stone to the "real thing" that will eventually come around - either a "good job" or successful self-employment.

It goes both ways - companies use their workers, workers use the companies. With the absence of pensions and unions, nobody has an incentive to maintain long term relationships, other than for a lack of better options.

[Bob Smith]

I think there are two discussions, which overlap, but aren't quite the same. One is the "gig" economy discussion, the other is the "decline of a job for life with a single employer" discussion. This gets at the second discussion - and isn't terribly surprising in that context.

[cm]

Bob: That's because it is first of all not clear what "gig" really means. One etymology source says the word emerged in the early 1900's in the bar/club/event musician scene. (This is the main context in which I heard the term when I was young - no that wasn't all the way back then.)

So "gig" means an engagement to perform services that is from the outset know to all parties to be short term, or even a single event.

The BLS seems to define "job" as an ongoing and recurring arrangement. (Which may also be short term, with a recurrence of every day, ever few days, etc.) By this definition, a job extends over a period, however short. But there is an overlap with "gig" - e.g. an engagement to perform at an event that goes on for a whole week.

So let's say if a limo driver or limo service is contracted to drive somebody on a specific day every week, for the duration of 1 or 2 months. Is that a gig, or a job, or a sequence of gigs?

When somebody is hired by a company for a period (or "permanent" which just means there is no set end date), that's definitely a job, but people may perceive it as a gig, i.e. "I'm serving here for the time being", but it's either not really what they want to do, or they know they will not retire from there.

[Kevin H]

The Pew research is problematic in that the "gig economy" trends are longer than the timescale they're talking about, and you probably want to look at the larger picture where employer tenure has dropped sharply from mid-century highs. There is also an extremely stark divide in tenure between gender and skill levels, where non-college educated males have seen steep declines in employer tenure after a peak in the early 1980s. This has been offset almost completely by an increase in employer tenure among female workers (with and without college degrees). This has the effect of masking a decline amongst a particular demographic in much the same way the overall labor force participation rate, masks a precipitous fall in male labor force participation over the same period.

This being said with the caveat that often the "gig economy" argument is often an updated stand-in for low-skill prime age male problems with attachment to the labor force overall.

[Kevin H]

Also of note, some accompanying US dataset research on employer tenure:

https://www.ebri.org/pdf/notespdf/EBRI_Notes_02_Feb15_Tenure-WBS.pdf

Long story short, men come out worse for the wear, women more than make up the difference.