Sticking With the Same Job Isn't Out of Style
It is widely believed that Americans change jobs a lot more than they used to. After years of digging through employment data in search of tectonic shifts, I have found that such widely held beliefs:
- Usually exaggerate the pace of change, by a lot.
- Nonetheless contain some, or even many, grains of truth.
So when I came across a reference on Twitter this week to a 2003 newspaper column claiming that "average job tenure" in the U.S. had fallen from 22.5 years in 1960 to 3.6 in 1996, I figured that couldn't be right. There are always lots of young people who have just started jobs, so to get to an average tenure of 22.5 years, you would also need lots of people who had been with the same employer for 30 or 40 years -- which in 1960 meant they would have held that job right through the Great Depression and World War II.
That sent me looking for the right number. The Bureau of Labor Statistics now measures job tenure every two years, and in 2014 the median was 4.6 years. The BLS also makes it easy to find job-tenure data back to 1983.
Getting the numbers from before then is harder, because the BLS doesn't want people to think they're really comparable to the numbers starting in 1983. And they aren't really comparable. Among other changes, before 1983 the BLS measured job tenure among all workers, including the self-employed, while since then it has just counted wage and salary workers. The longest-tenured occupation before 1983 was pretty much always self-employed farmer, which skews the pre-1983 numbers upward. Then again, there are a lot fewer self-employed farmers now than there were in the 1960s and 1970s, so it probably doesn't distort the numbers that much. I plowed forward, with help from an Employee Benefit Research Institute report and a JSTOR account, and harvested the pre-1983 numbers from back issues of the Monthly Labor Review, a BLS publication. Here's what I found, with a few years omitted to make the chart chronologically consistent and easier to read:
The BLS didn't measure job tenure in 1960, but it did in 1963, and the median tenure then for workers 14 and older was 4.6 years. That is a lot less than 22.5 years. Also, remarkably, it is exactly the same as the median tenure for workers 16 and older (the age shifted from 14 to 16 in the late 1960s) in 2014. One shouldn't make too much of this exactness -- as noted, the numbers aren't perfectly comparable. But still, it does undercut the argument that everybody is changing jobs more often when job tenure is now at or near its all-time peak.
Still, as you can see from the chart, there are some interesting things going on in the details. Men's median job tenure declined a lot from 1963 (5.7 years) to 2000 (3.8 years), and while it has risen some since (to 4.7 years in 2014), it's still well below its 1963 high. But as more women pursued careers, their median job tenure rose, as did their influence on overall median tenure. The trajectory of the overall tenure number is thus the result of composition effects that mask some of the changes going on.
Now that's a significant decline, from 12.8 years in 1983 to 8.2 in 2014. The numbers do indicate that, even in 1983, lifetime careers weren't the norm. But a 36 percent decline in median job tenure means lots of middle-aged men with much less stable careers. (If you want to see tenure changes for other age groups, check out the aforementioned EBRI report.)
Another way of looking at tenure is to measure the percentage of people who have been at the same job for a particular length of time. Here's the percentage of employed men 16 and older (14 and over in the 1960s) who've worked for the same employer for 10 or more years. I couldn't find this data for the 1980s, so the chart covers a different set of years than the others, and the 2000s are overrepresented.
That shows a decline too, albeit one that has reversed a bit since 2006. So yes, American men switch jobs more often than they used to, even if the change hasn't been as dramatic as many people seem to think.
1966, 1998, 2002, 2006, 2010 and 2012.
This time I omitted 1968 instead of 1966 because the number I was looking for wasn't reported in 1968.
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
To contact the author of this story:
Justin Fox at firstname.lastname@example.org
To contact the editor responsible for this story:
James Greiff at email@example.com